training_ver2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: keras in /usr/local/lib/python3.5/dist-packages\r\n",
      "Requirement already satisfied: scipy>=0.14 in /usr/local/lib/python3.5/dist-packages (from keras)\r\n",
      "Requirement already satisfied: pyyaml in /usr/local/lib/python3.5/dist-packages (from keras)\r\n",
      "Requirement already satisfied: six>=1.9.0 in /usr/local/lib/python3.5/dist-packages (from keras)\r\n",
      "Requirement already satisfied: numpy>=1.9.1 in /usr/local/lib/python3.5/dist-packages (from keras)\r\n"
     ]
    }
   ],
   "source": [
    "!pip install keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: cython in /usr/local/lib/python3.5/dist-packages\r\n"
     ]
    }
   ],
   "source": [
    "!pip install cython"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.append('Mask_RCNN')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import sys\n",
    "import random\n",
    "import math\n",
    "import re\n",
    "import time\n",
    "\n",
    "import tensorflow as tf\n",
    "config = tf.ConfigProto()\n",
    "config.gpu_options.allow_growth = True\n",
    "sess = tf.InteractiveSession(config=config)\n",
    "#sess = tf.Session(config=config)\n",
    "\n",
    "from keras import backend as K\n",
    "K.set_session(sess)\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import csv\n",
    "import imageio\n",
    "import skimage\n",
    "from config import Config\n",
    "from skimage.morphology import label\n",
    "from skimage.feature import canny\n",
    "from skimage import exposure\n",
    "from keras.callbacks import Callback\n",
    "from skimage.morphology import binary_closing, binary_opening, disk, binary_dilation\n",
    "from scipy.ndimage.morphology import binary_fill_holes\n",
    "from sklearn.externals import joblib\n",
    "from skimage.transform import PiecewiseAffineTransform, warp\n",
    "from skimage.morphology import watershed\n",
    "from skimage.filters import sobel\n",
    "from preparation import vgg_feature_extractor\n",
    "from random import *\n",
    "\n",
    "import gc; gc.enable()\n",
    "import utils\n",
    "import model as modellib\n",
    "import visualize\n",
    "from model import log"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's download the mask_rcnn trained on COCO dataset to initialize the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.options.display.max_rows=100\n",
    "\n",
    "# Root directory of the project\n",
    "ROOT_DIR = os.getcwd()\n",
    "\n",
    "# Directory to save logs and trained model\n",
    "MODEL_DIR = os.path.join(ROOT_DIR, \"logs\")\n",
    "\n",
    "# Local path to trained weights file\n",
    "COCO_MODEL_PATH = os.path.join(ROOT_DIR, \"mask_rcnn_coco.h5\")\n",
    "# Download COCO trained weights from Releases if needed\n",
    "if not os.path.exists(COCO_MODEL_PATH):\n",
    "    utils.download_trained_weights(COCO_MODEL_PATH)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will now define the monitor that will help keep track on the training process.\n",
    "With this you will be able to see your charts in the `charts` tab"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will load the metadata and divide them in train and valid splits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAIN_PATH = '../stage1_train'\n",
    "TEST_PATH = '../stage1_test'\n",
    "META_DATA_PATH = '../stage1_metadata.csv'\n",
    "BG_DATA_PATH = '../stage1_backgroud_foreground_classes.csv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ImageId</th>\n",
       "      <th>is_train</th>\n",
       "      <th>file_path_image</th>\n",
       "      <th>vgg_features_clusters</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ef6634efb46567d87b811be786b18c4cd0e2cda23d79b6...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/ef6634efb46567d87b811be786b18c...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>351771edfc5db5665ded8aa4940257276b6526663c76e3...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/351771edfc5db5665ded8aa4940257...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>785555c0cbb49dad835635217085287a8cc61c27d26f0e...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/785555c0cbb49dad83563521708528...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7ba20aa731cc21af74a8d940254176cbad1bdc44f240b5...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/7ba20aa731cc21af74a8d940254176...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1f...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/43cf6b2ec0b0745ac2b87b4d8780f6...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0b2e702f90aee4fff2bc6e4326308d50cf04701082e718...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/0b2e702f90aee4fff2bc6e4326308d...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>61dc249314d7b965eb4561ec739eab9b0f60af55c97b25...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/61dc249314d7b965eb4561ec739eab...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>28d33efef218392e79e385906deb88055d94b65ad217de...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/28d33efef218392e79e385906deb88...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>4596961c789d3b41916492918797724fe75128239fefc5...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/4596961c789d3b4191649291879772...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1ec74a26e772966df764e063f1391109a60d803cff9d15...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/1ec74a26e772966df764e063f13911...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>5c235b945b25b9905b9b0429ce59f1db51d0d0c7d48c2c...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/5c235b945b25b9905b9b0429ce59f1...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>TCGA-21-5784-01Z-00-DX1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-21-5784-01Z-00-DX1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>f952cc65376009cfad8249e53b9b2c0daaa3553e897096...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f952cc65376009cfad8249e53b9b2c...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>3b0709483b1e86449cc355bb797e841117ba178c6ae1ed...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/3b0709483b1e86449cc355bb797e84...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>3582166ee20755856adf4882a8bfacb616fce424791160...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/3582166ee20755856adf4882a8bfac...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>e5384c905e9879cb6e8ff5250fb03155bc1db035d8dde4...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/e5384c905e9879cb6e8ff5250fb031...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>a6593632dcbbe4c9e9429a9cec573d26fd8c91a47d554d...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a6593632dcbbe4c9e9429a9cec573d...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>TCGA-HE-7130-01Z-00-DX1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-HE-7130-01Z-00-DX1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>3bfd6bb152310f93daa6f4e1867c10572946e874b3a30c...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/3bfd6bb152310f93daa6f4e1867c10...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>305a8baaf726d7c9e695bff31d3a6a61445999a4732f0a...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/305a8baaf726d7c9e695bff31d3a6a...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>d7d12a2acc47a94961aeb56fd56e8a0873016af75f5dd1...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/d7d12a2acc47a94961aeb56fd56e8a...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>a90401357d50e1376354ae6e5f56a2e4dff3fdb5a4e8d5...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a90401357d50e1376354ae6e5f56a2...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>8fdc34509a0c3721f7b5e235c8a93e1f553343aa17ad10...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/8fdc34509a0c3721f7b5e235c8a93e...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>4d4ebfcae4374165ea6ae7c7e18fd0ba5014c3c860ee24...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/4d4ebfcae4374165ea6ae7c7e18fd0...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>bbce7ebc40323a0eff6574d0c3842f50f907f55fbfb46c...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/bbce7ebc40323a0eff6574d0c3842f...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>4948e3dfe0483c2198b9616612071ead5474dffd223071...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/4948e3dfe0483c2198b9616612071e...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>cc88627344305b9a9b07f8bd042cb074c7a834c13de67f...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/cc88627344305b9a9b07f8bd042cb0...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>88d5a03f8ecd459f076a06e0d5035149193bfdd727c309...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/88d5a03f8ecd459f076a06e0d50351...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>2dd3356f2dcf470aec4003800744dfec6490e75d88011e...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/2dd3356f2dcf470aec4003800744df...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>abbfff07379bceb69dba41dad8b0db5eb80cc8baf3d4af...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/abbfff07379bceb69dba41dad8b0db...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>6b72b61b80060a9e79a4747f9c5d5af135af9db466681c...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/6b72b61b80060a9e79a4747f9c5d5a...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>a1777737270c5f96c4523dff76e4097756f8f7d4c9d59b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a1777737270c5f96c4523dff76e409...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>f81ca7ee25e733ff37240c34c8e3044d9937bb0166e315...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f81ca7ee25e733ff37240c34c8e304...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>df9a4212ecb67bb4e58eba62f293b91f9d6f1dde73e38f...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/df9a4212ecb67bb4e58eba62f293b9...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>547ef286ee5f4e5dce533e982e6992ada67b7d727fdd3c...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/547ef286ee5f4e5dce533e982e6992...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>72b18a405555ad491721e29454e5cd325055ce81a9e785...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/72b18a405555ad491721e29454e5cd...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>751f421d322940d6efe3bd570a66ecda16d08a1b90bc32...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/751f421d322940d6efe3bd570a66ec...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f487cc82271cf84b4414552aa8b0a9...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>TCGA-RD-A8N9-01A-01-TS1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-RD-A8N9-01A-01-TS1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>06350c7cc618be442c15706db7a68e91f313758d224de4...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/06350c7cc618be442c15706db7a68e...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>TCGA-50-5931-01Z-00-DX1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-50-5931-01Z-00-DX1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>7aae06bc4558829473071defec0b7ab3bfa9c5005548a1...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/7aae06bc4558829473071defec0b7a...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>f9ac03b0344ce8c48bc058448541f9211a1e5f4c94fdaf...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f9ac03b0344ce8c48bc058448541f9...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>5ba4facefc949c920d7054813a3e846b000969da2ed860...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/5ba4facefc949c920d7054813a3e84...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>a6e81120d1cb9f71f8a25f90a5d56c4b714a642fc496a7...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a6e81120d1cb9f71f8a25f90a5d56c...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>6fc83b33896f58a4a067d8fdcf51f15d4ae9be05d8c381...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/6fc83b33896f58a4a067d8fdcf51f1...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>bf4a61bb81589c9a67e3343408befd3e135af5e88b50c1...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/bf4a61bb81589c9a67e3343408befd...</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>40bcdad218ac5f0885fc247d88fcad9f729f55c81c79d2...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/40bcdad218ac5f0885fc247d88fcad...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>b4de1e3eec159d8af1bd5447696f8996c31709edaf33e2...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/b4de1e3eec159d8af1bd5447696f89...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>573a657d5e5fcb9574a758b0ab34b09c79d7ba374ceb71...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/573a657d5e5fcb9574a758b0ab34b0...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>1e8408fbb1619e7a0bcdd0bcd21fae57e7cb1f297d4c79...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/1e8408fbb1619e7a0bcdd0bcd21fae...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/b6edad733399c83c8eb7a59c0d37b5...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>8f94a80b95a881d0efdec36affc915dca9609f4cba8134...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/8f94a80b95a881d0efdec36affc915...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>bb8ebf465c968a5f6f715de5d9e2e664afd1bcaa533e0e...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/bb8ebf465c968a5f6f715de5d9e2e6...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>8de0b1a2e8f614af29fe5fafeaa5bdf55e6b3e65edf363...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/8de0b1a2e8f614af29fe5fafeaa5bd...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>337b6eed0726f07531cd467cd62b6676c31a8c9e716bdb...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/337b6eed0726f07531cd467cd62b66...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>bde3727f3a9e8b2b58f383ebc762b2157eb50cdbff23e6...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/bde3727f3a9e8b2b58f383ebc762b2...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>a3a65623e079af7988b0c1cf1e54041003c6d730c91ecf...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a3a65623e079af7988b0c1cf1e5404...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>62057502c387145ed4f8f7f0d5e5bedcb72d3bcec15fa7...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/62057502c387145ed4f8f7f0d5e5be...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>309ba76b12ecb5ce28b99f3445b2b5dc54c0564c3c0e24...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/309ba76b12ecb5ce28b99f3445b2b5...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>449fe932622db3b49366a260ddd20077219f96fb2dc0f9...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/449fe932622db3b49366a260ddd200...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>b24ea5c268469a95ed155eeaf809e36030b78a2eb530a0...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/b24ea5c268469a95ed155eeaf809e3...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>c96109cbebcf206f20035cbde414e43872074eee8d839b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/c96109cbebcf206f20035cbde414e4...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>e856511ac1c34d24320eb7c56c05a4a3340d06667b4f5b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/e856511ac1c34d24320eb7c56c05a4...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>64</th>\n",
       "      <td>7b5987a24dd57325e82812371b3f4df7edc528e0526754...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/7b5987a24dd57325e82812371b3f4d...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>65</th>\n",
       "      <td>1f9e429c12f4477221b5b855a5f494fda2ef6d064ff75b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/1f9e429c12f4477221b5b855a5f494...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>648636ee314d7bdba3ab2fc0fe49a863de35c3e2caf619...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/648636ee314d7bdba3ab2fc0fe49a8...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>f73e37957c74f554be132986f38b6f1d75339f636dfe2b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f73e37957c74f554be132986f38b6f...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>623cf6987b3fac8f384c09f40d98c5e739c097aa9a9627...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/623cf6987b3fac8f384c09f40d98c5...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>8bef203fce625e4d8c89dca728158be4662dfdfdcd4dc7...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/8bef203fce625e4d8c89dca728158b...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>245b995878370ef4ea977568b2b67f93d4ecaa9308761b...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/245b995878370ef4ea977568b2b67f...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>57bd029b19c1b382bef9db3ac14f13ea85e36a6053b92e...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/57bd029b19c1b382bef9db3ac14f13...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>93cfd412c7de5210bbd262ec3a602cfea65072e9272e9f...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/93cfd412c7de5210bbd262ec3a602c...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>da5f98f2b8a64eee735a398de48ed42cd31bf17a6063db...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/da5f98f2b8a64eee735a398de48ed4...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>ce37f6dd0615d45e66e41a8f2ed6fbc0bbe3103a290394...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/ce37f6dd0615d45e66e41a8f2ed6fb...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>449a9c32e53a37c8a86e01c199155c8da3958b631088e1...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/449a9c32e53a37c8a86e01c199155c...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>20468e8779c43e089dc0ff30f25e6cf3872d5aa6a0fdad...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/20468e8779c43e089dc0ff30f25e6c...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>2bf594e9d06f78b4b79d7ffb395497a0a91126b6b0d710...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/2bf594e9d06f78b4b79d7ffb395497...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>TCGA-G9-6362-01Z-00-DX1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-G9-6362-01Z-00-DX1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>f43169e3d8b4f71e687945b9e72cbfdfe2e40e68842568...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f43169e3d8b4f71e687945b9e72cbf...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>866a8cba7bfe1ea73e383d6cf492e53752579140c8b833...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/866a8cba7bfe1ea73e383d6cf492e5...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>62570c4ff1c5ab6d9d383aba9f25e604768520b4266afd...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/62570c4ff1c5ab6d9d383aba9f25e6...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82</th>\n",
       "      <td>12f89395ad5d21491ab9cec137e247652451d283064773...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/12f89395ad5d21491ab9cec137e247...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>f728de04267283f0b4daab9a840e7433b2c6034baf195f...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f728de04267283f0b4daab9a840e74...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>08ae2741df2f5ac815c0f272a8c532b5167ee853be9b93...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/08ae2741df2f5ac815c0f272a8c532...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>58406ed8ef944831c413c3424dc2b07e59aef13eb1ff16...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/58406ed8ef944831c413c3424dc2b0...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>516a0e20327d6dfcedcf57e3056115e4fb29cdf4cb3490...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/516a0e20327d6dfcedcf57e3056115...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>TCGA-B0-5698-01Z-00-DX1</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/TCGA-B0-5698-01Z-00-DX1/images...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>a246bcaa64af48ee5ca181cd594c0fc43466e7614406eb...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/a246bcaa64af48ee5ca181cd594c0f...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>89be66f88612aae541f5843abcd9c015832b5d6c54a281...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/89be66f88612aae541f5843abcd9c0...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>8f6597cd978c060378177df76e554d0578b97eab471e23...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/8f6597cd978c060378177df76e554d...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>aa58ba4512955771b4f9b459cb4e6a8adb71d11cd6cae6...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/aa58ba4512955771b4f9b459cb4e6a...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>ff3407842ada5bc18be79ae453e5bdaa1b68afc842fc22...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/ff3407842ada5bc18be79ae453e5bd...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>edd36ed822e7ed760ff73e0524df22aa5bf5c565efcdc6...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/edd36ed822e7ed760ff73e0524df22...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>3ebd2ab34ba86e515feb79ffdeb7fc303a074a98ba3994...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/3ebd2ab34ba86e515feb79ffdeb7fc...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>f26f4c2c70c38fe12e00d5a814d5116691f2ca54890812...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/f26f4c2c70c38fe12e00d5a814d511...</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>1b6044e4858a9b7cee9b0028d8e54fbc8fb72e6c4424ab...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/1b6044e4858a9b7cee9b0028d8e54f...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>e66f25e175abab08ecb4e5f6859db64a211e0ddffb262d...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/e66f25e175abab08ecb4e5f6859db6...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>d7db360fabfce9828559a21f6bffff589ae868e0dc6101...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/d7db360fabfce9828559a21f6bffff...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>c04fa1a74a980d790ba6f3e595fd9851f14370bb71c7cb...</td>\n",
       "      <td>1</td>\n",
       "      <td>../stage1_train/c04fa1a74a980d790ba6f3e595fd98...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              ImageId  is_train  \\\n",
       "0   ef6634efb46567d87b811be786b18c4cd0e2cda23d79b6...         1   \n",
       "1   351771edfc5db5665ded8aa4940257276b6526663c76e3...         1   \n",
       "2   785555c0cbb49dad835635217085287a8cc61c27d26f0e...         1   \n",
       "3   7ba20aa731cc21af74a8d940254176cbad1bdc44f240b5...         1   \n",
       "4   43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1f...         1   \n",
       "5   0b2e702f90aee4fff2bc6e4326308d50cf04701082e718...         1   \n",
       "6   61dc249314d7b965eb4561ec739eab9b0f60af55c97b25...         1   \n",
       "7   28d33efef218392e79e385906deb88055d94b65ad217de...         1   \n",
       "8   4596961c789d3b41916492918797724fe75128239fefc5...         1   \n",
       "9   1ec74a26e772966df764e063f1391109a60d803cff9d15...         1   \n",
       "10  5c235b945b25b9905b9b0429ce59f1db51d0d0c7d48c2c...         1   \n",
       "11                            TCGA-21-5784-01Z-00-DX1         1   \n",
       "12  f952cc65376009cfad8249e53b9b2c0daaa3553e897096...         1   \n",
       "13  3b0709483b1e86449cc355bb797e841117ba178c6ae1ed...         1   \n",
       "14  3582166ee20755856adf4882a8bfacb616fce424791160...         1   \n",
       "15  e5384c905e9879cb6e8ff5250fb03155bc1db035d8dde4...         1   \n",
       "16  a6593632dcbbe4c9e9429a9cec573d26fd8c91a47d554d...         1   \n",
       "17                            TCGA-HE-7130-01Z-00-DX1         1   \n",
       "18  3bfd6bb152310f93daa6f4e1867c10572946e874b3a30c...         1   \n",
       "19  305a8baaf726d7c9e695bff31d3a6a61445999a4732f0a...         1   \n",
       "20  d7d12a2acc47a94961aeb56fd56e8a0873016af75f5dd1...         1   \n",
       "21  a90401357d50e1376354ae6e5f56a2e4dff3fdb5a4e8d5...         1   \n",
       "22  8fdc34509a0c3721f7b5e235c8a93e1f553343aa17ad10...         1   \n",
       "23  4d4ebfcae4374165ea6ae7c7e18fd0ba5014c3c860ee24...         1   \n",
       "24  bbce7ebc40323a0eff6574d0c3842f50f907f55fbfb46c...         1   \n",
       "25  4948e3dfe0483c2198b9616612071ead5474dffd223071...         1   \n",
       "26  cc88627344305b9a9b07f8bd042cb074c7a834c13de67f...         1   \n",
       "27  88d5a03f8ecd459f076a06e0d5035149193bfdd727c309...         1   \n",
       "28  2dd3356f2dcf470aec4003800744dfec6490e75d88011e...         1   \n",
       "29  abbfff07379bceb69dba41dad8b0db5eb80cc8baf3d4af...         1   \n",
       "30  6b72b61b80060a9e79a4747f9c5d5af135af9db466681c...         1   \n",
       "31  a1777737270c5f96c4523dff76e4097756f8f7d4c9d59b...         1   \n",
       "32  f81ca7ee25e733ff37240c34c8e3044d9937bb0166e315...         1   \n",
       "33  df9a4212ecb67bb4e58eba62f293b91f9d6f1dde73e38f...         1   \n",
       "34  547ef286ee5f4e5dce533e982e6992ada67b7d727fdd3c...         1   \n",
       "35  72b18a405555ad491721e29454e5cd325055ce81a9e785...         1   \n",
       "36  751f421d322940d6efe3bd570a66ecda16d08a1b90bc32...         1   \n",
       "37  f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8...         1   \n",
       "38                            TCGA-RD-A8N9-01A-01-TS1         1   \n",
       "39  06350c7cc618be442c15706db7a68e91f313758d224de4...         1   \n",
       "40                            TCGA-50-5931-01Z-00-DX1         1   \n",
       "41  7aae06bc4558829473071defec0b7ab3bfa9c5005548a1...         1   \n",
       "42  f9ac03b0344ce8c48bc058448541f9211a1e5f4c94fdaf...         1   \n",
       "43  5ba4facefc949c920d7054813a3e846b000969da2ed860...         1   \n",
       "44  a6e81120d1cb9f71f8a25f90a5d56c4b714a642fc496a7...         1   \n",
       "45  6fc83b33896f58a4a067d8fdcf51f15d4ae9be05d8c381...         1   \n",
       "46  bf4a61bb81589c9a67e3343408befd3e135af5e88b50c1...         1   \n",
       "47  40bcdad218ac5f0885fc247d88fcad9f729f55c81c79d2...         1   \n",
       "48  b4de1e3eec159d8af1bd5447696f8996c31709edaf33e2...         1   \n",
       "49  573a657d5e5fcb9574a758b0ab34b09c79d7ba374ceb71...         1   \n",
       "50  1e8408fbb1619e7a0bcdd0bcd21fae57e7cb1f297d4c79...         1   \n",
       "51  b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39...         1   \n",
       "52  8f94a80b95a881d0efdec36affc915dca9609f4cba8134...         1   \n",
       "53  bb8ebf465c968a5f6f715de5d9e2e664afd1bcaa533e0e...         1   \n",
       "54  8de0b1a2e8f614af29fe5fafeaa5bdf55e6b3e65edf363...         1   \n",
       "55  337b6eed0726f07531cd467cd62b6676c31a8c9e716bdb...         1   \n",
       "56  bde3727f3a9e8b2b58f383ebc762b2157eb50cdbff23e6...         1   \n",
       "57  a3a65623e079af7988b0c1cf1e54041003c6d730c91ecf...         1   \n",
       "58  62057502c387145ed4f8f7f0d5e5bedcb72d3bcec15fa7...         1   \n",
       "59  309ba76b12ecb5ce28b99f3445b2b5dc54c0564c3c0e24...         1   \n",
       "60  449fe932622db3b49366a260ddd20077219f96fb2dc0f9...         1   \n",
       "61  b24ea5c268469a95ed155eeaf809e36030b78a2eb530a0...         1   \n",
       "62  c96109cbebcf206f20035cbde414e43872074eee8d839b...         1   \n",
       "63  e856511ac1c34d24320eb7c56c05a4a3340d06667b4f5b...         1   \n",
       "64  7b5987a24dd57325e82812371b3f4df7edc528e0526754...         1   \n",
       "65  1f9e429c12f4477221b5b855a5f494fda2ef6d064ff75b...         1   \n",
       "66  648636ee314d7bdba3ab2fc0fe49a863de35c3e2caf619...         1   \n",
       "67  f73e37957c74f554be132986f38b6f1d75339f636dfe2b...         1   \n",
       "68  623cf6987b3fac8f384c09f40d98c5e739c097aa9a9627...         1   \n",
       "69  8bef203fce625e4d8c89dca728158be4662dfdfdcd4dc7...         1   \n",
       "70  245b995878370ef4ea977568b2b67f93d4ecaa9308761b...         1   \n",
       "71  57bd029b19c1b382bef9db3ac14f13ea85e36a6053b92e...         1   \n",
       "72  93cfd412c7de5210bbd262ec3a602cfea65072e9272e9f...         1   \n",
       "73  da5f98f2b8a64eee735a398de48ed42cd31bf17a6063db...         1   \n",
       "74  ce37f6dd0615d45e66e41a8f2ed6fbc0bbe3103a290394...         1   \n",
       "75  449a9c32e53a37c8a86e01c199155c8da3958b631088e1...         1   \n",
       "76  20468e8779c43e089dc0ff30f25e6cf3872d5aa6a0fdad...         1   \n",
       "77  2bf594e9d06f78b4b79d7ffb395497a0a91126b6b0d710...         1   \n",
       "78                            TCGA-G9-6362-01Z-00-DX1         1   \n",
       "79  f43169e3d8b4f71e687945b9e72cbfdfe2e40e68842568...         1   \n",
       "80  866a8cba7bfe1ea73e383d6cf492e53752579140c8b833...         1   \n",
       "81  62570c4ff1c5ab6d9d383aba9f25e604768520b4266afd...         1   \n",
       "82  12f89395ad5d21491ab9cec137e247652451d283064773...         1   \n",
       "83  f728de04267283f0b4daab9a840e7433b2c6034baf195f...         1   \n",
       "84  08ae2741df2f5ac815c0f272a8c532b5167ee853be9b93...         1   \n",
       "85  58406ed8ef944831c413c3424dc2b07e59aef13eb1ff16...         1   \n",
       "86  516a0e20327d6dfcedcf57e3056115e4fb29cdf4cb3490...         1   \n",
       "87                            TCGA-B0-5698-01Z-00-DX1         1   \n",
       "88  a246bcaa64af48ee5ca181cd594c0fc43466e7614406eb...         1   \n",
       "89  89be66f88612aae541f5843abcd9c015832b5d6c54a281...         1   \n",
       "90  8f6597cd978c060378177df76e554d0578b97eab471e23...         1   \n",
       "91  aa58ba4512955771b4f9b459cb4e6a8adb71d11cd6cae6...         1   \n",
       "92  ff3407842ada5bc18be79ae453e5bdaa1b68afc842fc22...         1   \n",
       "93  edd36ed822e7ed760ff73e0524df22aa5bf5c565efcdc6...         1   \n",
       "94  3ebd2ab34ba86e515feb79ffdeb7fc303a074a98ba3994...         1   \n",
       "95  f26f4c2c70c38fe12e00d5a814d5116691f2ca54890812...         1   \n",
       "96  1b6044e4858a9b7cee9b0028d8e54fbc8fb72e6c4424ab...         1   \n",
       "97  e66f25e175abab08ecb4e5f6859db64a211e0ddffb262d...         1   \n",
       "98  d7db360fabfce9828559a21f6bffff589ae868e0dc6101...         1   \n",
       "99  c04fa1a74a980d790ba6f3e595fd9851f14370bb71c7cb...         1   \n",
       "\n",
       "                                      file_path_image  vgg_features_clusters  \n",
       "0   ../stage1_train/ef6634efb46567d87b811be786b18c...                      1  \n",
       "1   ../stage1_train/351771edfc5db5665ded8aa4940257...                      6  \n",
       "2   ../stage1_train/785555c0cbb49dad83563521708528...                      4  \n",
       "3   ../stage1_train/7ba20aa731cc21af74a8d940254176...                      1  \n",
       "4   ../stage1_train/43cf6b2ec0b0745ac2b87b4d8780f6...                      3  \n",
       "5   ../stage1_train/0b2e702f90aee4fff2bc6e4326308d...                      4  \n",
       "6   ../stage1_train/61dc249314d7b965eb4561ec739eab...                      6  \n",
       "7   ../stage1_train/28d33efef218392e79e385906deb88...                      0  \n",
       "8   ../stage1_train/4596961c789d3b4191649291879772...                      6  \n",
       "9   ../stage1_train/1ec74a26e772966df764e063f13911...                      9  \n",
       "10  ../stage1_train/5c235b945b25b9905b9b0429ce59f1...                      0  \n",
       "11  ../stage1_train/TCGA-21-5784-01Z-00-DX1/images...                      7  \n",
       "12  ../stage1_train/f952cc65376009cfad8249e53b9b2c...                      2  \n",
       "13  ../stage1_train/3b0709483b1e86449cc355bb797e84...                      9  \n",
       "14  ../stage1_train/3582166ee20755856adf4882a8bfac...                      5  \n",
       "15  ../stage1_train/e5384c905e9879cb6e8ff5250fb031...                      0  \n",
       "16  ../stage1_train/a6593632dcbbe4c9e9429a9cec573d...                      5  \n",
       "17  ../stage1_train/TCGA-HE-7130-01Z-00-DX1/images...                      7  \n",
       "18  ../stage1_train/3bfd6bb152310f93daa6f4e1867c10...                      1  \n",
       "19  ../stage1_train/305a8baaf726d7c9e695bff31d3a6a...                      5  \n",
       "20  ../stage1_train/d7d12a2acc47a94961aeb56fd56e8a...                      0  \n",
       "21  ../stage1_train/a90401357d50e1376354ae6e5f56a2...                      9  \n",
       "22  ../stage1_train/8fdc34509a0c3721f7b5e235c8a93e...                      1  \n",
       "23  ../stage1_train/4d4ebfcae4374165ea6ae7c7e18fd0...                      6  \n",
       "24  ../stage1_train/bbce7ebc40323a0eff6574d0c3842f...                      0  \n",
       "25  ../stage1_train/4948e3dfe0483c2198b9616612071e...                      0  \n",
       "26  ../stage1_train/cc88627344305b9a9b07f8bd042cb0...                      6  \n",
       "27  ../stage1_train/88d5a03f8ecd459f076a06e0d50351...                      5  \n",
       "28  ../stage1_train/2dd3356f2dcf470aec4003800744df...                      2  \n",
       "29  ../stage1_train/abbfff07379bceb69dba41dad8b0db...                      1  \n",
       "30  ../stage1_train/6b72b61b80060a9e79a4747f9c5d5a...                      6  \n",
       "31  ../stage1_train/a1777737270c5f96c4523dff76e409...                      0  \n",
       "32  ../stage1_train/f81ca7ee25e733ff37240c34c8e304...                      5  \n",
       "33  ../stage1_train/df9a4212ecb67bb4e58eba62f293b9...                      1  \n",
       "34  ../stage1_train/547ef286ee5f4e5dce533e982e6992...                      6  \n",
       "35  ../stage1_train/72b18a405555ad491721e29454e5cd...                      4  \n",
       "36  ../stage1_train/751f421d322940d6efe3bd570a66ec...                      1  \n",
       "37  ../stage1_train/f487cc82271cf84b4414552aa8b0a9...                      6  \n",
       "38  ../stage1_train/TCGA-RD-A8N9-01A-01-TS1/images...                      7  \n",
       "39  ../stage1_train/06350c7cc618be442c15706db7a68e...                      4  \n",
       "40  ../stage1_train/TCGA-50-5931-01Z-00-DX1/images...                      7  \n",
       "41  ../stage1_train/7aae06bc4558829473071defec0b7a...                      1  \n",
       "42  ../stage1_train/f9ac03b0344ce8c48bc058448541f9...                      1  \n",
       "43  ../stage1_train/5ba4facefc949c920d7054813a3e84...                      0  \n",
       "44  ../stage1_train/a6e81120d1cb9f71f8a25f90a5d56c...                      0  \n",
       "45  ../stage1_train/6fc83b33896f58a4a067d8fdcf51f1...                      0  \n",
       "46  ../stage1_train/bf4a61bb81589c9a67e3343408befd...                      9  \n",
       "47  ../stage1_train/40bcdad218ac5f0885fc247d88fcad...                      4  \n",
       "48  ../stage1_train/b4de1e3eec159d8af1bd5447696f89...                      0  \n",
       "49  ../stage1_train/573a657d5e5fcb9574a758b0ab34b0...                      1  \n",
       "50  ../stage1_train/1e8408fbb1619e7a0bcdd0bcd21fae...                      4  \n",
       "51  ../stage1_train/b6edad733399c83c8eb7a59c0d37b5...                      0  \n",
       "52  ../stage1_train/8f94a80b95a881d0efdec36affc915...                      5  \n",
       "53  ../stage1_train/bb8ebf465c968a5f6f715de5d9e2e6...                      6  \n",
       "54  ../stage1_train/8de0b1a2e8f614af29fe5fafeaa5bd...                      4  \n",
       "55  ../stage1_train/337b6eed0726f07531cd467cd62b66...                      2  \n",
       "56  ../stage1_train/bde3727f3a9e8b2b58f383ebc762b2...                      6  \n",
       "57  ../stage1_train/a3a65623e079af7988b0c1cf1e5404...                      4  \n",
       "58  ../stage1_train/62057502c387145ed4f8f7f0d5e5be...                      6  \n",
       "59  ../stage1_train/309ba76b12ecb5ce28b99f3445b2b5...                      4  \n",
       "60  ../stage1_train/449fe932622db3b49366a260ddd200...                      6  \n",
       "61  ../stage1_train/b24ea5c268469a95ed155eeaf809e3...                      3  \n",
       "62  ../stage1_train/c96109cbebcf206f20035cbde414e4...                      0  \n",
       "63  ../stage1_train/e856511ac1c34d24320eb7c56c05a4...                      5  \n",
       "64  ../stage1_train/7b5987a24dd57325e82812371b3f4d...                      6  \n",
       "65  ../stage1_train/1f9e429c12f4477221b5b855a5f494...                      1  \n",
       "66  ../stage1_train/648636ee314d7bdba3ab2fc0fe49a8...                      1  \n",
       "67  ../stage1_train/f73e37957c74f554be132986f38b6f...                      1  \n",
       "68  ../stage1_train/623cf6987b3fac8f384c09f40d98c5...                      1  \n",
       "69  ../stage1_train/8bef203fce625e4d8c89dca728158b...                      5  \n",
       "70  ../stage1_train/245b995878370ef4ea977568b2b67f...                      6  \n",
       "71  ../stage1_train/57bd029b19c1b382bef9db3ac14f13...                      2  \n",
       "72  ../stage1_train/93cfd412c7de5210bbd262ec3a602c...                      4  \n",
       "73  ../stage1_train/da5f98f2b8a64eee735a398de48ed4...                      6  \n",
       "74  ../stage1_train/ce37f6dd0615d45e66e41a8f2ed6fb...                      4  \n",
       "75  ../stage1_train/449a9c32e53a37c8a86e01c199155c...                      3  \n",
       "76  ../stage1_train/20468e8779c43e089dc0ff30f25e6c...                      5  \n",
       "77  ../stage1_train/2bf594e9d06f78b4b79d7ffb395497...                      4  \n",
       "78  ../stage1_train/TCGA-G9-6362-01Z-00-DX1/images...                      7  \n",
       "79  ../stage1_train/f43169e3d8b4f71e687945b9e72cbf...                      1  \n",
       "80  ../stage1_train/866a8cba7bfe1ea73e383d6cf492e5...                      4  \n",
       "81  ../stage1_train/62570c4ff1c5ab6d9d383aba9f25e6...                      4  \n",
       "82  ../stage1_train/12f89395ad5d21491ab9cec137e247...                      1  \n",
       "83  ../stage1_train/f728de04267283f0b4daab9a840e74...                      1  \n",
       "84  ../stage1_train/08ae2741df2f5ac815c0f272a8c532...                      4  \n",
       "85  ../stage1_train/58406ed8ef944831c413c3424dc2b0...                      0  \n",
       "86  ../stage1_train/516a0e20327d6dfcedcf57e3056115...                      0  \n",
       "87  ../stage1_train/TCGA-B0-5698-01Z-00-DX1/images...                      7  \n",
       "88  ../stage1_train/a246bcaa64af48ee5ca181cd594c0f...                      6  \n",
       "89  ../stage1_train/89be66f88612aae541f5843abcd9c0...                      0  \n",
       "90  ../stage1_train/8f6597cd978c060378177df76e554d...                      4  \n",
       "91  ../stage1_train/aa58ba4512955771b4f9b459cb4e6a...                      2  \n",
       "92  ../stage1_train/ff3407842ada5bc18be79ae453e5bd...                      3  \n",
       "93  ../stage1_train/edd36ed822e7ed760ff73e0524df22...                      6  \n",
       "94  ../stage1_train/3ebd2ab34ba86e515feb79ffdeb7fc...                      6  \n",
       "95  ../stage1_train/f26f4c2c70c38fe12e00d5a814d511...                      6  \n",
       "96  ../stage1_train/1b6044e4858a9b7cee9b0028d8e54f...                      1  \n",
       "97  ../stage1_train/e66f25e175abab08ecb4e5f6859db6...                      4  \n",
       "98  ../stage1_train/d7db360fabfce9828559a21f6bffff...                      0  \n",
       "99  ../stage1_train/c04fa1a74a980d790ba6f3e595fd98...                      1  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "if not os.path.exists(META_DATA_PATH):\n",
    "    print('Generating metadata.csv')\n",
    "    \n",
    "    stage1_metadata = pd.DataFrame()\n",
    "    \n",
    "    train_image_ids = os.listdir(TRAIN_PATH) # ImageId\n",
    "    train_is_train = [1] * len(train_image_ids)\n",
    "    \n",
    "    train_file_path = []\n",
    "    for p in train_image_ids:\n",
    "        if p[:4] == 'TCGA':\n",
    "            train_file_path.append(TRAIN_PATH + '/' + p + '/images/' + p + '.tif')\n",
    "        else:\n",
    "            train_file_path.append(TRAIN_PATH + '/' + p + '/images/' + p + '.png')\n",
    "    \n",
    "    test_image_ids = os.listdir(TEST_PATH) # ImageId\n",
    "    test_is_train = [0] * len(test_image_ids)\n",
    "    test_file_path = [TEST_PATH + '/' + p + '/images/' + p + '.png' for p in test_image_ids]\n",
    "    \n",
    "    stage1_metadata['ImageId'] = np.concatenate([train_image_ids, test_image_ids], axis=0)\n",
    "    stage1_metadata['is_train'] = np.concatenate([train_is_train, test_is_train], axis=0)\n",
    "    stage1_metadata['file_path_image'] = np.concatenate([train_file_path, test_file_path], axis=0)\n",
    "    \n",
    "    ve = vgg_feature_extractor(n_clusters=10)\n",
    "    train_vgg_features_clusters = ve.get_vgg_clusters(stage1_metadata[stage1_metadata.is_train == 1].reset_index(drop=True), is_train=True)\n",
    "    test_vgg_features_clusters =ve.get_vgg_clusters(stage1_metadata[stage1_metadata.is_train == 0].reset_index(drop=True), is_train=False)\n",
    "    \n",
    "    stage1_metadata['vgg_features_clusters'] = np.concatenate([train_vgg_features_clusters, test_vgg_features_clusters], axis=0)\n",
    "    \n",
    "    stage1_metadata.to_csv(META_DATA_PATH, index=False)\n",
    "    meta = stage1_metadata\n",
    "    \n",
    "    del vgg_feature_extractor; gc.collect()\n",
    "else:\n",
    "    meta = pd.read_csv(META_DATA_PATH)\n",
    "    \n",
    "\n",
    "meta.head(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>filename</th>\n",
       "      <th>foreground</th>\n",
       "      <th>background</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>665</th>\n",
       "      <td>589f86dee5b480a88dd4f77eeaffe2c4d70aefdf879a40...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>666</th>\n",
       "      <td>442c4eb0185698fe7d148c108a46f74abd399aecda2f4f...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>667</th>\n",
       "      <td>709e094e39629a9ca21e187f007b331074694e443db402...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>668</th>\n",
       "      <td>6b0ac2ab04c09dced54058ec504a4947f8ecd5727dfca7...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>669</th>\n",
       "      <td>87ea72894f6534b28e740cc34cf5c9eb75d0d8902687fc...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>670</th>\n",
       "      <td>b0e35e06b85da49bfe3ea737711a72b551a6add446e30e...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>671</th>\n",
       "      <td>f952cc65376009cfad8249e53b9b2c0daaa3553e897096...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>672</th>\n",
       "      <td>6d327ab4f0e3604fa6e9b8041c7e6db86ab809890d886c...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>673</th>\n",
       "      <td>e49fc2b4f1f39d481a6525225ab3f688be5c87f5688445...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>674</th>\n",
       "      <td>a90401357d50e1376354ae6e5f56a2e4dff3fdb5a4e8d5...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>675</th>\n",
       "      <td>a7f6194ddbeaefb1da571226a97785d09ccafc5893ce3c...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>676</th>\n",
       "      <td>af576e8ec3a8d0b57eb6a311299e9e4fd2047970d3dd9d...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>677</th>\n",
       "      <td>a31deaf0ac279d5f34fb2eca80cc2abce6ef30bd64e7ac...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>678</th>\n",
       "      <td>136000dc18fa6def2d6c98d4d0b2084d13c22eaffe82e2...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>679</th>\n",
       "      <td>9fb32aba1c2fd53273dca9abefac944ba747f578da82df...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>680</th>\n",
       "      <td>6c67b78e8164801059375ed9a607f61e67a7ae347e92e3...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>681</th>\n",
       "      <td>4d4f254f3b8b4408d661df3735591554b2f6587ce19529...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>682</th>\n",
       "      <td>5ef4442e5b8b0b4cf824b61be4050dfd793d846e0a6800...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>683</th>\n",
       "      <td>13f2bec0a24c70345372febb14c4352877b1b6c1b01896...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>684</th>\n",
       "      <td>f7e5dcfc9c93183c668c5a4ab028d5faad54fb54298711...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685</th>\n",
       "      <td>65c8527c16a016191118e8adc3d307fe3a73d37cbe0559...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>686</th>\n",
       "      <td>2dd3356f2dcf470aec4003800744dfec6490e75d88011e...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>687</th>\n",
       "      <td>45f059cf21d85ecfce0eb93260516f1e2443d210e9a52f...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>688</th>\n",
       "      <td>472b1c5ff988dadc209faea92499bc07f305208dbda29d...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>689</th>\n",
       "      <td>4e23ecf65040f86420e0201134e538951acdeda84fbb27...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>690</th>\n",
       "      <td>a08166d91d2cca263d2dd52764dc25c9c582b7a5ece2b8...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>691</th>\n",
       "      <td>c0152b1a260e71f9823d17f4fbb4bf7020d5dce62b4a12...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>692</th>\n",
       "      <td>0d3640c1f1b80f24e94cc9a5f3e1d9e8db7bf6af7d4aba...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>693</th>\n",
       "      <td>b909aa8f6f4bec37c3fb6ff5a85d166162d07983506fcc...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>694</th>\n",
       "      <td>4d14a3629b6af6de86d850be236b833a7bfcbf6d8665fd...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>695</th>\n",
       "      <td>5c6eb9a47852754d4e45eceb9a696c64c7cfe304afc5ea...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>696</th>\n",
       "      <td>3934a094e8537841e973342c7f8880606f7a2712b14930...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>697</th>\n",
       "      <td>a7f767ca9770b160f234780e172aeb35a50830ba10dc49...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>698</th>\n",
       "      <td>a5fe0b7412dd152c41f7afc34ffdf276d4261b6942fa6d...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>699</th>\n",
       "      <td>953211bcc0192e2298087d30e708dba68def9e0c13a3ff...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>700</th>\n",
       "      <td>24a20f2a529cede5695df2422a3986505b5826bb10b107...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>701</th>\n",
       "      <td>cbff60361ded0570e5d50429a1aa51d81471819bc9b383...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>702</th>\n",
       "      <td>4ae4f936a9ade472764dad80f60f7168e4be067aa66ce9...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>703</th>\n",
       "      <td>2e2d29fc44444a85049b162eb359a523dec108ccd5bd75...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>704</th>\n",
       "      <td>ae570a676961482848b5097038ef5e407df7a66a8e1c9b...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>705</th>\n",
       "      <td>ad473063dab4bf4f2461d9a99a9c0166d4871f156516d9...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>706</th>\n",
       "      <td>01d44a26f6680c42ba94c9bc6339228579a95d0e2695b1...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>707</th>\n",
       "      <td>6ab24e7e1f6c9fdd371c5edae1bbb20abeeb976811f8ab...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>708</th>\n",
       "      <td>708eb41a3fc8f2b6cd1f529cdf38dc4ad5d5f00ad30bdc...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>709</th>\n",
       "      <td>1e488c42eb1a54a3e8412b1f12cde530f950f238d71078...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>710</th>\n",
       "      <td>74a7785530687a11ecd073e772f90912d9967d02407a19...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>711</th>\n",
       "      <td>673baf65ae5c571d6be452eb41e79ef3fc2eb3fd238e62...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>712</th>\n",
       "      <td>8e507d58f4c27cd2a82bee79fe27b069befd62a46fdaed...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>713</th>\n",
       "      <td>4cbd6c37f3a55a538d759d440344c287cac66260d3047a...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>714</th>\n",
       "      <td>ebc18868864ad075548cc1784f4f9a237bb98335f9645e...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>715</th>\n",
       "      <td>abd8dde78f8d37b68b28da67459371ed65f0a575523e94...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>716</th>\n",
       "      <td>2246874c8b5ba218d01ad8153a201ad4660195f3e4c65d...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>717</th>\n",
       "      <td>815524d88283ba10ad597b87aa1967671db776df8004a0...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>718</th>\n",
       "      <td>3b0709483b1e86449cc355bb797e841117ba178c6ae1ed...</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>719</th>\n",
       "      <td>8f94a80b95a881d0efdec36affc915dca9609f4cba8134...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>720</th>\n",
       "      <td>4217e25defac94ff465157d53f5a24b8a14045b763d860...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>721</th>\n",
       "      <td>7f38885521586fc6011bef1314a9fb2aa1e4935bd581b2...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>722</th>\n",
       "      <td>c395870ad9f5a3ae651b50efab9b20c3e6b9aea15d4c73...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>723</th>\n",
       "      <td>4e07a653352b30bb95b60ebc6c57afbc7215716224af73...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>724</th>\n",
       "      <td>76a372bfd3fad3ea30cb163b560e52607a8281f5b04248...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>725</th>\n",
       "      <td>1b44d22643830cd4f23c9deadb0bd499fb392fb2cd9526...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>726</th>\n",
       "      <td>091944f1d2611c916b98c020bd066667e33f4639159b2a...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>727</th>\n",
       "      <td>8d05fb18ee0cda107d56735cafa6197a31884e0a5092dc...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>728</th>\n",
       "      <td>1a11552569160f0b1ea10bedbd628ce6c14f29edec5092...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>729</th>\n",
       "      <td>08275a5b1c2dfcd739e8c4888a5ee2d29f83eccfa75185...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>730</th>\n",
       "      <td>5d58600efa0c2667ec85595bf456a54e2bd6e6e9a5c0df...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>731</th>\n",
       "      <td>54793624413c7d0e048173f7aeee85de3277f7e8d47c82...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>732</th>\n",
       "      <td>5e263abff938acba1c0cff698261c7c00c23d7376e3cea...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>733</th>\n",
       "      <td>3594684b9ea0e16196f498815508f8d364d55fea2933a2...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>734</th>\n",
       "      <td>2a1a294e21d76efd0399e4eb321b45f44f7510911acd92...</td>\n",
       "      <td>black</td>\n",
       "      <td>white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>735</th>\n",
       "      <td>TCGA-18-5592-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>736</th>\n",
       "      <td>TCGA-21-5784-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>737</th>\n",
       "      <td>TCGA-21-5786-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>738</th>\n",
       "      <td>TCGA-38-6178-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>739</th>\n",
       "      <td>TCGA-49-4488-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>740</th>\n",
       "      <td>TCGA-50-5931-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>TCGA-A7-A13E-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>TCGA-A7-A13F-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>TCGA-AR-A1AK-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>TCGA-AR-A1AS-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>TCGA-AY-A8YK-01A-01-TS1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>746</th>\n",
       "      <td>TCGA-B0-5698-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>747</th>\n",
       "      <td>TCGA-B0-5710-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>748</th>\n",
       "      <td>TCGA-B0-5711-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>749</th>\n",
       "      <td>TCGA-CH-5767-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>750</th>\n",
       "      <td>TCGA-DK-A2I6-01A-01-TS1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>751</th>\n",
       "      <td>TCGA-E2-A14V-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>752</th>\n",
       "      <td>TCGA-E2-A1B5-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>753</th>\n",
       "      <td>TCGA-G2-A2EK-01A-02-TSB.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>754</th>\n",
       "      <td>TCGA-G9-6336-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>755</th>\n",
       "      <td>TCGA-G9-6348-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>756</th>\n",
       "      <td>TCGA-G9-6356-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>757</th>\n",
       "      <td>TCGA-G9-6362-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>758</th>\n",
       "      <td>TCGA-G9-6363-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>759</th>\n",
       "      <td>TCGA-HE-7128-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760</th>\n",
       "      <td>TCGA-HE-7129-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761</th>\n",
       "      <td>TCGA-HE-7130-01Z-00-DX1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762</th>\n",
       "      <td>TCGA-KB-A93J-01A-01-TS1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>763</th>\n",
       "      <td>TCGA-NH-A8F7-01A-01-TS1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>764</th>\n",
       "      <td>TCGA-RD-A8N9-01A-01-TS1.tif</td>\n",
       "      <td>purple</td>\n",
       "      <td>purple</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              filename foreground background\n",
       "665  589f86dee5b480a88dd4f77eeaffe2c4d70aefdf879a40...     purple     purple\n",
       "666  442c4eb0185698fe7d148c108a46f74abd399aecda2f4f...     purple     purple\n",
       "667  709e094e39629a9ca21e187f007b331074694e443db402...     purple     purple\n",
       "668  6b0ac2ab04c09dced54058ec504a4947f8ecd5727dfca7...     purple     purple\n",
       "669  87ea72894f6534b28e740cc34cf5c9eb75d0d8902687fc...     purple     purple\n",
       "670  b0e35e06b85da49bfe3ea737711a72b551a6add446e30e...     purple     purple\n",
       "671  f952cc65376009cfad8249e53b9b2c0daaa3553e897096...     purple     purple\n",
       "672  6d327ab4f0e3604fa6e9b8041c7e6db86ab809890d886c...     purple     purple\n",
       "673  e49fc2b4f1f39d481a6525225ab3f688be5c87f5688445...     purple     purple\n",
       "674  a90401357d50e1376354ae6e5f56a2e4dff3fdb5a4e8d5...     purple     purple\n",
       "675  a7f6194ddbeaefb1da571226a97785d09ccafc5893ce3c...     purple     purple\n",
       "676  af576e8ec3a8d0b57eb6a311299e9e4fd2047970d3dd9d...     purple     purple\n",
       "677  a31deaf0ac279d5f34fb2eca80cc2abce6ef30bd64e7ac...     purple     purple\n",
       "678  136000dc18fa6def2d6c98d4d0b2084d13c22eaffe82e2...     purple     purple\n",
       "679  9fb32aba1c2fd53273dca9abefac944ba747f578da82df...     purple     purple\n",
       "680  6c67b78e8164801059375ed9a607f61e67a7ae347e92e3...     purple     purple\n",
       "681  4d4f254f3b8b4408d661df3735591554b2f6587ce19529...     purple     purple\n",
       "682  5ef4442e5b8b0b4cf824b61be4050dfd793d846e0a6800...     purple     purple\n",
       "683  13f2bec0a24c70345372febb14c4352877b1b6c1b01896...     purple     purple\n",
       "684  f7e5dcfc9c93183c668c5a4ab028d5faad54fb54298711...     purple     purple\n",
       "685  65c8527c16a016191118e8adc3d307fe3a73d37cbe0559...     purple     purple\n",
       "686  2dd3356f2dcf470aec4003800744dfec6490e75d88011e...     purple     purple\n",
       "687  45f059cf21d85ecfce0eb93260516f1e2443d210e9a52f...     purple     purple\n",
       "688  472b1c5ff988dadc209faea92499bc07f305208dbda29d...     purple     purple\n",
       "689  4e23ecf65040f86420e0201134e538951acdeda84fbb27...     purple     purple\n",
       "690  a08166d91d2cca263d2dd52764dc25c9c582b7a5ece2b8...     purple     purple\n",
       "691  c0152b1a260e71f9823d17f4fbb4bf7020d5dce62b4a12...     purple     purple\n",
       "692  0d3640c1f1b80f24e94cc9a5f3e1d9e8db7bf6af7d4aba...     purple     purple\n",
       "693  b909aa8f6f4bec37c3fb6ff5a85d166162d07983506fcc...     purple     purple\n",
       "694  4d14a3629b6af6de86d850be236b833a7bfcbf6d8665fd...     purple     purple\n",
       "695  5c6eb9a47852754d4e45eceb9a696c64c7cfe304afc5ea...     purple     purple\n",
       "696  3934a094e8537841e973342c7f8880606f7a2712b14930...     purple     purple\n",
       "697  a7f767ca9770b160f234780e172aeb35a50830ba10dc49...     purple     purple\n",
       "698  a5fe0b7412dd152c41f7afc34ffdf276d4261b6942fa6d...     purple     purple\n",
       "699  953211bcc0192e2298087d30e708dba68def9e0c13a3ff...     purple     purple\n",
       "700  24a20f2a529cede5695df2422a3986505b5826bb10b107...     purple     purple\n",
       "701  cbff60361ded0570e5d50429a1aa51d81471819bc9b383...     purple     purple\n",
       "702  4ae4f936a9ade472764dad80f60f7168e4be067aa66ce9...     purple     purple\n",
       "703  2e2d29fc44444a85049b162eb359a523dec108ccd5bd75...     purple     purple\n",
       "704  ae570a676961482848b5097038ef5e407df7a66a8e1c9b...     purple     purple\n",
       "705  ad473063dab4bf4f2461d9a99a9c0166d4871f156516d9...     purple     purple\n",
       "706  01d44a26f6680c42ba94c9bc6339228579a95d0e2695b1...     purple     purple\n",
       "707  6ab24e7e1f6c9fdd371c5edae1bbb20abeeb976811f8ab...     purple     purple\n",
       "708  708eb41a3fc8f2b6cd1f529cdf38dc4ad5d5f00ad30bdc...     purple     purple\n",
       "709  1e488c42eb1a54a3e8412b1f12cde530f950f238d71078...     purple     purple\n",
       "710  74a7785530687a11ecd073e772f90912d9967d02407a19...     purple     purple\n",
       "711  673baf65ae5c571d6be452eb41e79ef3fc2eb3fd238e62...     purple     purple\n",
       "712  8e507d58f4c27cd2a82bee79fe27b069befd62a46fdaed...     purple     purple\n",
       "713  4cbd6c37f3a55a538d759d440344c287cac66260d3047a...     purple     purple\n",
       "714  ebc18868864ad075548cc1784f4f9a237bb98335f9645e...     purple     purple\n",
       "715  abd8dde78f8d37b68b28da67459371ed65f0a575523e94...     purple     purple\n",
       "716  2246874c8b5ba218d01ad8153a201ad4660195f3e4c65d...     purple     purple\n",
       "717  815524d88283ba10ad597b87aa1967671db776df8004a0...     purple     purple\n",
       "718  3b0709483b1e86449cc355bb797e841117ba178c6ae1ed...     purple     purple\n",
       "719  8f94a80b95a881d0efdec36affc915dca9609f4cba8134...      black      white\n",
       "720  4217e25defac94ff465157d53f5a24b8a14045b763d860...      black      white\n",
       "721  7f38885521586fc6011bef1314a9fb2aa1e4935bd581b2...      black      white\n",
       "722  c395870ad9f5a3ae651b50efab9b20c3e6b9aea15d4c73...      black      white\n",
       "723  4e07a653352b30bb95b60ebc6c57afbc7215716224af73...      black      white\n",
       "724  76a372bfd3fad3ea30cb163b560e52607a8281f5b04248...      black      white\n",
       "725  1b44d22643830cd4f23c9deadb0bd499fb392fb2cd9526...      black      white\n",
       "726  091944f1d2611c916b98c020bd066667e33f4639159b2a...      black      white\n",
       "727  8d05fb18ee0cda107d56735cafa6197a31884e0a5092dc...      black      white\n",
       "728  1a11552569160f0b1ea10bedbd628ce6c14f29edec5092...      black      white\n",
       "729  08275a5b1c2dfcd739e8c4888a5ee2d29f83eccfa75185...      black      white\n",
       "730  5d58600efa0c2667ec85595bf456a54e2bd6e6e9a5c0df...      black      white\n",
       "731  54793624413c7d0e048173f7aeee85de3277f7e8d47c82...      black      white\n",
       "732  5e263abff938acba1c0cff698261c7c00c23d7376e3cea...      black      white\n",
       "733  3594684b9ea0e16196f498815508f8d364d55fea2933a2...      black      white\n",
       "734  2a1a294e21d76efd0399e4eb321b45f44f7510911acd92...      black      white\n",
       "735                        TCGA-18-5592-01Z-00-DX1.tif     purple     purple\n",
       "736                        TCGA-21-5784-01Z-00-DX1.tif     purple     purple\n",
       "737                        TCGA-21-5786-01Z-00-DX1.tif     purple     purple\n",
       "738                        TCGA-38-6178-01Z-00-DX1.tif     purple     purple\n",
       "739                        TCGA-49-4488-01Z-00-DX1.tif     purple     purple\n",
       "740                        TCGA-50-5931-01Z-00-DX1.tif     purple     purple\n",
       "741                        TCGA-A7-A13E-01Z-00-DX1.tif     purple     purple\n",
       "742                        TCGA-A7-A13F-01Z-00-DX1.tif     purple     purple\n",
       "743                        TCGA-AR-A1AK-01Z-00-DX1.tif     purple     purple\n",
       "744                        TCGA-AR-A1AS-01Z-00-DX1.tif     purple     purple\n",
       "745                        TCGA-AY-A8YK-01A-01-TS1.tif     purple     purple\n",
       "746                        TCGA-B0-5698-01Z-00-DX1.tif     purple     purple\n",
       "747                        TCGA-B0-5710-01Z-00-DX1.tif     purple     purple\n",
       "748                        TCGA-B0-5711-01Z-00-DX1.tif     purple     purple\n",
       "749                        TCGA-CH-5767-01Z-00-DX1.tif     purple     purple\n",
       "750                        TCGA-DK-A2I6-01A-01-TS1.tif     purple     purple\n",
       "751                        TCGA-E2-A14V-01Z-00-DX1.tif     purple     purple\n",
       "752                        TCGA-E2-A1B5-01Z-00-DX1.tif     purple     purple\n",
       "753                        TCGA-G2-A2EK-01A-02-TSB.tif     purple     purple\n",
       "754                        TCGA-G9-6336-01Z-00-DX1.tif     purple     purple\n",
       "755                        TCGA-G9-6348-01Z-00-DX1.tif     purple     purple\n",
       "756                        TCGA-G9-6356-01Z-00-DX1.tif     purple     purple\n",
       "757                        TCGA-G9-6362-01Z-00-DX1.tif     purple     purple\n",
       "758                        TCGA-G9-6363-01Z-00-DX1.tif     purple     purple\n",
       "759                        TCGA-HE-7128-01Z-00-DX1.tif     purple     purple\n",
       "760                        TCGA-HE-7129-01Z-00-DX1.tif     purple     purple\n",
       "761                        TCGA-HE-7130-01Z-00-DX1.tif     purple     purple\n",
       "762                        TCGA-KB-A93J-01A-01-TS1.tif     purple     purple\n",
       "763                        TCGA-NH-A8F7-01A-01-TS1.tif     purple     purple\n",
       "764                        TCGA-RD-A8N9-01A-01-TS1.tif     purple     purple"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "back_ground_df = pd.read_csv(BG_DATA_PATH)\n",
    "back_ground_df.tail(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Image ID in background data counts:\n",
      " True    759\n",
      "Name: ImageId, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "back_ground_df['ImageId'] = [fn.replace('.png', '') if fn[-4:] == '.png' else fn.replace('.tif', '') for fn in back_ground_df.filename]\n",
    "print('Image ID in background data counts:\\n', meta.ImageId.isin(back_ground_df.ImageId).value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "black     599\n",
       "purple    101\n",
       "white      57\n",
       "yellow      8\n",
       "Name: background, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "back_ground_df.background.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "white     599\n",
       "purple    150\n",
       "black      16\n",
       "Name: foreground, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "back_ground_df.foreground.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "black_white      599\n",
       "purple_purple    101\n",
       "white_purple      41\n",
       "white_black       16\n",
       "yellow_purple      8\n",
       "Name: composed_bf_ground, dtype: int64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "back_ground_df['composed_bf_ground'] = back_ground_df.background.astype(str) + '_' + back_ground_df.foreground.astype(str)\n",
    "back_ground_df.composed_bf_ground.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### How to deal with it:\n",
    "1. yellow_purple, white_black -> reverse to black_white\n",
    "2. white_purple, purple_purple -> change to white black. (the foreground seems darker in these cases)\n",
    "3. Separate into two groups: 1. + 2."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>filename</th>\n",
       "      <th>foreground</th>\n",
       "      <th>background</th>\n",
       "      <th>composed_bf_ground</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ImageId</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d1173f5150dde8ff38cf</th>\n",
       "      <td>d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d1...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4596961c789d3b41916492918797724fe75128239fefc516c3ee75322b7926f0</th>\n",
       "      <td>4596961c789d3b41916492918797724fe75128239fefc5...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39ff843884d84f61dee1</th>\n",
       "      <td>b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>831218e6a1a54b23d4be56c5799854e7eb978811b89215319dc138900bd563e6</th>\n",
       "      <td>831218e6a1a54b23d4be56c5799854e7eb978811b89215...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845bdc01b2b3ee2299229</th>\n",
       "      <td>b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16ebf46babe4656a794e</th>\n",
       "      <td>4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1fcefa42d166191e84c6</th>\n",
       "      <td>43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1f...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8bc639d4121c1672ff7</th>\n",
       "      <td>f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bf7691b0a79811fa068b7408cbce636a73f01ef9e971a95da1a2d96df73782b6</th>\n",
       "      <td>bf7691b0a79811fa068b7408cbce636a73f01ef9e971a9...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>df33b11184427e05c8a450f921586685975fe975f57315e686a0f26fddb93db1</th>\n",
       "      <td>df33b11184427e05c8a450f921586685975fe975f57315...</td>\n",
       "      <td>white</td>\n",
       "      <td>black</td>\n",
       "      <td>black_white</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                             filename  \\\n",
       "ImageId                                                                                                 \n",
       "d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d11...  d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d1...   \n",
       "4596961c789d3b41916492918797724fe75128239fefc51...  4596961c789d3b41916492918797724fe75128239fefc5...   \n",
       "b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39f...  b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39...   \n",
       "831218e6a1a54b23d4be56c5799854e7eb978811b892153...  831218e6a1a54b23d4be56c5799854e7eb978811b89215...   \n",
       "b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845b...  b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845...   \n",
       "4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16e...  4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16...   \n",
       "43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1fc...  43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1f...   \n",
       "f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8b...  f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8...   \n",
       "bf7691b0a79811fa068b7408cbce636a73f01ef9e971a95...  bf7691b0a79811fa068b7408cbce636a73f01ef9e971a9...   \n",
       "df33b11184427e05c8a450f921586685975fe975f57315e...  df33b11184427e05c8a450f921586685975fe975f57315...   \n",
       "\n",
       "                                                   foreground background  \\\n",
       "ImageId                                                                    \n",
       "d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d11...      white      black   \n",
       "4596961c789d3b41916492918797724fe75128239fefc51...      white      black   \n",
       "b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39f...      white      black   \n",
       "831218e6a1a54b23d4be56c5799854e7eb978811b892153...      white      black   \n",
       "b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845b...      white      black   \n",
       "4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16e...      white      black   \n",
       "43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1fc...      white      black   \n",
       "f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8b...      white      black   \n",
       "bf7691b0a79811fa068b7408cbce636a73f01ef9e971a95...      white      black   \n",
       "df33b11184427e05c8a450f921586685975fe975f57315e...      white      black   \n",
       "\n",
       "                                                   composed_bf_ground  \n",
       "ImageId                                                                \n",
       "d32ea6d318626ca14a967d0c1ad3218aebfe636624a8d11...        black_white  \n",
       "4596961c789d3b41916492918797724fe75128239fefc51...        black_white  \n",
       "b6edad733399c83c8eb7a59c0d37b54e10cc0d59894e39f...        black_white  \n",
       "831218e6a1a54b23d4be56c5799854e7eb978811b892153...        black_white  \n",
       "b1e3aeb0c56261c17eb71c747d116057b8da7e8c8a6845b...        black_white  \n",
       "4a424e0cb845cf6fd4d9fe62875552c7b89a4e0276cf16e...        black_white  \n",
       "43cf6b2ec0b0745ac2b87b4d8780f62e9050d3f5d50a1fc...        black_white  \n",
       "f487cc82271cf84b4414552aa8b0a9d82d902451ebe8e8b...        black_white  \n",
       "bf7691b0a79811fa068b7408cbce636a73f01ef9e971a95...        black_white  \n",
       "df33b11184427e05c8a450f921586685975fe975f57315e...        black_white  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "back_ground_df = back_ground_df.set_index('ImageId')\n",
    "back_ground_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "black_white      594\n",
       "purple_purple    100\n",
       "white_purple      41\n",
       "white_black       16\n",
       "yellow_purple      8\n",
       "Name: bg_type, dtype: int64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta['bg_type'] = meta.ImageId.map(back_ground_df.composed_bf_ground)\n",
    "meta['bg_type'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "black_white      541\n",
       "purple_purple     99\n",
       "white_purple      38\n",
       "white_black       16\n",
       "Name: bg_type, dtype: int64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta[meta['is_train']==1].bg_type.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "black_white      53\n",
       "yellow_purple     8\n",
       "white_purple      3\n",
       "purple_purple     1\n",
       "Name: bg_type, dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta[meta['is_train']==0].bg_type.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2    594\n",
       "1    149\n",
       "3     16\n",
       "Name: process_type, dtype: int64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta['process_type'] = meta['bg_type'].map({\n",
    "    'black_white': 2, # do nothing\n",
    "    'white_black': 3, # pure reverse\n",
    "    'purple_purple': 1, # turns into gray and reverse\n",
    "    'yellow_purple': 1, # turns into gray and reverse\n",
    "    'white_purple': 1 # turns into gray and reverse\n",
    "})\n",
    "meta['process_type'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    759\n",
       "Name: group, dtype: int64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meta['group'] = meta['bg_type'].map({\n",
    "    'black_white': 0,\n",
    "    'white_black': 0,\n",
    "    'purple_purple': 0,\n",
    "    'yellow_purple': 0,\n",
    "    'white_purple': 0\n",
    "})\n",
    "meta['group'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_val_split(meta_train, val_size=0.1, random_state=719):\n",
    "    meta_train_groups = [meta_train[meta_train['group'] == group].reset_index(drop=True) for group in range(1)]\n",
    "    \n",
    "    meta_train_val_groups = []\n",
    "    \n",
    "    for train_group in meta_train_groups:\n",
    "        meta_train_val_groups\n",
    "        val = train_group.sample(frac=val_size, random_state=random_state)\n",
    "        tr = train_group[~train_group.index.isin(val.index)]\n",
    "        \n",
    "        meta_train_val_groups.append((tr, val))\n",
    "    \n",
    "    return meta_train_val_groups\n",
    "\n",
    "def train_split(meta_train):\n",
    "    meta_train_groups = [meta_train[meta_train['group'] == group].reset_index(drop=True) for group in range(2)]\n",
    "    return meta_train_groups"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "IS_DEVELOP = False\n",
    "if IS_DEVELOP:\n",
    "    VAL_SIZE = .3\n",
    "else:\n",
    "    VAL_SIZE = 0.01"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(687, 7) (7, 7)\n"
     ]
    }
   ],
   "source": [
    "meta_train = meta[meta['is_train']==1]\n",
    "meta_test = meta[meta['is_train']==0]\n",
    "meta_train_val_groups = train_val_split(meta_train, val_size=VAL_SIZE, random_state=519)\n",
    "\n",
    "for i in range(1):\n",
    "    print(meta_train_val_groups[i][0].shape, meta_train_val_groups[i][1].shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Mask-RCNN configuration\n",
    "We will now define parameters for our mask-rcnn. \n",
    "Tweak them to get better results!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DsbConfig(Config):\n",
    "\n",
    "    # Give the configuration a recognizable name\n",
    "    NAME = \"dsb\"\n",
    "      \n",
    "    LEARNING_RATE = 1e-2\n",
    "    \n",
    "    # If enabled, resizes instance masks to a smaller size to reduce\n",
    "    # memory load. Recommended when using high-resolution image\n",
    "    USE_MINI_MASK = True\n",
    "    MINI_MASK_SHAPE = (56, 56)  # (height, width) of the mini-mask\n",
    "    \n",
    "    # Train on 1 GPU and 8 images per GPU. Batch size is GPUs * images/GPU.\n",
    "    GPU_COUNT = 1\n",
    "    IMAGES_PER_GPU = 2\n",
    "    # Total number of steps (batches of samples) to yield from generator before declaring one epoch finished and starting the next epoch.\n",
    "    # typically be equal to the number of samples of your dataset divided by the batch size\n",
    "    STEPS_PER_EPOCH = 300\n",
    "    VALIDATION_STEPS = 70\n",
    "\n",
    "    # Number of classes (including background)\n",
    "    NUM_CLASSES = 1 + 1  # background + nucleis\n",
    "    IMAGE_MIN_DIM = 512\n",
    "    IMAGE_MAX_DIM = 512\n",
    "    IMAGE_PADDING = True  # currently, the False option is not supported\n",
    "    RPN_ANCHOR_SCALES = (8, 16, 32, 64, 128)  # anchor side in pixels, maybe add a 256?\n",
    "    # The strides of each layer of the FPN Pyramid. These values\n",
    "    # are based on a Resnet101 backbone.\n",
    "    BACKBONE_STRIDES = [4, 8, 16, 32, 64]\n",
    "    # How many anchors per image to use for RPN training\n",
    "    RPN_TRAIN_ANCHORS_PER_IMAGE = 320 #300\n",
    "    \n",
    "    # ROIs kept after non-maximum supression (training and inference)\n",
    "    POST_NMS_ROIS_TRAINING = 2000\n",
    "    POST_NMS_ROIS_INFERENCE = 2000\n",
    "    POOL_SIZE = 7\n",
    "    MASK_POOL_SIZE = 14 # 14\n",
    "    MASK_SHAPE = [28, 28] # 28\n",
    "    TRAIN_ROIS_PER_IMAGE = 512\n",
    "    RPN_DETECTION_MIN_CONFIDENCE = .6\n",
    "    RPN_NMS_THRESHOLD = 0.6 #0.7\n",
    "    MAX_GT_INSTANCES = 256\n",
    "    DETECTION_MAX_INSTANCES = 450 \n",
    "    # Minimum probability value to accept a detected instance\n",
    "    # ROIs below this threshold are skipped\n",
    "    DETECTION_MIN_CONFIDENCE = 0.67 # may be smaller?\n",
    "    # Non-maximum suppression threshold for detection\n",
    "    DETECTION_NMS_THRESHOLD = 0.3 # 0.3\n",
    "    \n",
    "    # DETECTION_MAX_INSTANCES=400, DETECTION_MIN_CONFIDENCE=.7 => Val loss: .21xx, LB: .424\n",
    "    # DETECTION_MAX_INSTANCES=450, DETECTION_MIN_CONFIDENCE=.65 => Val loss: .3, LB: .438\n",
    "    # DETECTION_MAX_INSTANCES=, DETECTION_MIN_CONFIDENCE= => Val loss: ., LB: .\n",
    "    \n",
    "    '''\n",
    "    0324 note: \n",
    "        RPN_NMS_THRESHOLD .7 -> .6 to generate more proposals at training,\n",
    "        DETECTION_MAX_INSTANCES 400 -> 450\n",
    "        DETECTION_MIN_CONFIDENCE .7\n",
    "        BACKBONE_STRIDES add `2` ..?\n",
    "    '''\n",
    "    \n",
    "    MEAN_PIXEL = np.array([0.,0.,0.])\n",
    "    \n",
    "    # Weight decay regularization\n",
    "    WEIGHT_DECAY = 0.0001\n",
    "    \n",
    "    # Min Box Size\n",
    "    MIN_BOX_SIZE = 36./(512.**2.) # (36/512**2) (delete objects smaller than 6*6 pixels)\n",
    "    BORDER_RATIO = 4. / 512. # smaller than 4 pixels width\\height at the border\n",
    "    BORDER_MIN_BOX_SIZE = 16./(512.**2.)\n",
    "  \n",
    "    LR_SHRINK_PERIOD = 10\n",
    "    \n",
    "config = DsbConfig()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Agument functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Agg_Func():\n",
    "    def __init__(self, funcs=[]):\n",
    "        self.funcs = funcs\n",
    "        \n",
    "    def __call__(self, x):\n",
    "        res = x\n",
    "        for f in self.funcs:\n",
    "            res = f(res)\n",
    "        return res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Augment():\n",
    "    def __init__(self, rotate=True, flip=True):\n",
    "        self.augment_funcs = []\n",
    "        \n",
    "        if flip:   \n",
    "            self.augment_funcs.extend([\n",
    "                Agg_Func(funcs=[]),\n",
    "                Agg_Func(funcs=[np.fliplr]),\n",
    "                Agg_Func(funcs=[np.flipud]),\n",
    "                Agg_Func(funcs=[np.fliplr, np.flipud]),\n",
    "            ])\n",
    "            \n",
    "            if rotate:\n",
    "                self.augment_funcs.extend([\n",
    "                    Agg_Func(funcs=[np.rot90]),\n",
    "                    Agg_Func(funcs=[np.rot90, np.fliplr]),\n",
    "                    Agg_Func(funcs=[np.rot90, np.flipud]),\n",
    "                    Agg_Func(funcs=[np.rot90, np.fliplr, np.flipud]),\n",
    "                ])\n",
    "        elif rotate:\n",
    "            self.augment_funcs.extend([\n",
    "                Agg_Func(funcs=[]),\n",
    "                Agg_Func(funcs=[np.rot90]),\n",
    "            ])\n",
    "    \n",
    "    def get_random_augement(self):\n",
    "        index = randrange(len(self.augment_funcs))\n",
    "        return self.augment_funcs[index]\n",
    "    \n",
    "    def apply_all_augment(self, img):\n",
    "        res = []\n",
    "        for f in self.augment_funcs:\n",
    "            res.append(f(img))\n",
    "        return res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dataset definition\n",
    "\n",
    "We need to define our datasets for the data science bowl competition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DsbDataset(utils.Dataset):\n",
    "\n",
    "    '''\n",
    "    process_info is a image-id <=> process type mapping pd.series\n",
    "    '''\n",
    "    def load_dataset(self, process_info, train_mode=True, augment=False):\n",
    "        self.add_class(\"dsb\", 1, \"nuclei\")\n",
    "        self.process_info = process_info\n",
    "        self.augmenter = Augment()\n",
    "        self.augment = augment\n",
    "        \n",
    "        if train_mode:\n",
    "            directory = dsb_dir\n",
    "        else:\n",
    "            directory = test_dir\n",
    "            \n",
    "        ids = []\n",
    "        for i, id in enumerate(process_info.ImageId.values):\n",
    "            image_dir = os.path.join(directory, id)\n",
    "            self.add_image(\"dsb\", image_id=i, path=image_dir)\n",
    "            ids.append(i)\n",
    "        \n",
    "        self.process_info.reset_index(drop=True, inplace=True)\n",
    "        #print(self.process_info)\n",
    "        self.process_info['id'] = ids\n",
    "        self.process_info.set_index('id')\n",
    "        self.process_info = self.process_info['process_type']\n",
    "        #print(self.process_info)\n",
    "        \n",
    "    def load_image(self, image_id, non_zero=None):\n",
    "        info = self.image_info[image_id]\n",
    "        path = info['path']\n",
    "        image_name = os.listdir(os.path.join(path, 'images'))\n",
    "        image_path = os.path.join(path, 'images', image_name[0])\n",
    "        image = imageio.imread(image_path)\n",
    "        \n",
    "        #print(self.process_info[image_id])\n",
    "        if image_path[-4:] == '.tif':\n",
    "            self.offset = (randrange(488), randrange(488))\n",
    "            image = image[self.offset[0]:self.offset[0]+512, self.offset[1]:self.offset[1]+512, :]\n",
    "            #image = skimage.transform.resize(image, (512,512),  preserve_range=True)\n",
    "        \n",
    "        if image.shape[2] != 3:\n",
    "            image = image[:,:,:3]\n",
    "\n",
    "        image = self.preprocess(image, process_type=self.process_info[image_id])\n",
    "        image = image.astype('float32')\n",
    "        \n",
    "        if self.augment:\n",
    "            self.augment_func = self.augmenter.get_random_augement()\n",
    "            image = self.augment_func(image)\n",
    "            \n",
    "        return image\n",
    "\n",
    "    def image_reference(self, image_id):\n",
    "        info = self.image_info[image_id]\n",
    "        if info[\"source\"] == \"shapes\":\n",
    "            return info[\"shapes\"]\n",
    "        else:\n",
    "            super(self.__class__).image_reference(self, image_id)\n",
    "\n",
    "    def load_mask(self, image_id):\n",
    "        info = self.image_info[image_id]\n",
    "        path = info['path']\n",
    "        mask_dir = os.path.join(path, 'masks')\n",
    "        mask_names = os.listdir(mask_dir)\n",
    "        count = len(mask_names)\n",
    "        mask = []\n",
    "        \n",
    "        for i, el in enumerate(mask_names):\n",
    "            msk_path = os.path.join(mask_dir, el)\n",
    "            msk = imageio.imread(msk_path)\n",
    "            if np.sum(msk) == 0:\n",
    "                print('invalid mask')\n",
    "                continue\n",
    "                \n",
    "            if 'TCGA' in mask_dir:\n",
    "                #msk = skimage.transform.resize(msk, (512,512), preserve_range=True)\n",
    "                msk = msk[self.offset[0]:self.offset[0]+512, self.offset[1]:self.offset[1]+512]\n",
    "                if np.sum(msk) == 0:\n",
    "                    continue\n",
    "                \n",
    "            msk = msk.astype('float32')/255.\n",
    "            \n",
    "            if len(msk.shape) > 2:\n",
    "                msk = np.sum(msk, axis=2)\n",
    "                \n",
    "            mask.append(msk)\n",
    "            \n",
    "        try:\n",
    "            mask = np.array(mask)\n",
    "        except:\n",
    "            for i, m in enumerate(mask):\n",
    "                if len(m.shape) > 2:\n",
    "                    print(mask_names[i])\n",
    "                    \n",
    "        mask[mask > 0.] = 1.\n",
    "        mask = np.transpose(mask, (1,2,0))\n",
    "        if self.augment:\n",
    "            mask = self.augment_func(mask) # self.augment_func was decided when loading image before loading masks\n",
    "            \n",
    "        occlusion = np.logical_not(mask[:, :, -1]).astype(np.uint8)\n",
    "        count = mask.shape[2]\n",
    "        for i in range(count-2, -1, -1):\n",
    "            mask[:, :, i] = mask[:, :, i] * occlusion\n",
    "            occlusion = np.logical_and(occlusion, np.logical_not(mask[:, :, i]))\n",
    "        class_ids = [self.class_names.index('nuclei') for s in range(count)]\n",
    "        class_ids = np.asarray(class_ids)\n",
    "        \n",
    "        return mask, class_ids.astype(np.int32)\n",
    "    \n",
    "    def preprocess(self, img, process_type=0):\n",
    "        \n",
    "        if process_type == 0: # grey-rgb-grey\n",
    "            gray = skimage.color.rgb2gray(img.astype('uint8'))\n",
    "            img = skimage.color.gray2rgb(gray)\n",
    "            img *= 255.\n",
    "        elif process_type == 1:  # grey-rgb-grey-reverse\n",
    "            gray = skimage.color.rgb2gray(img.astype('uint8'))\n",
    "            img = skimage.color.gray2rgb(gray)\n",
    "            img *= 255.\n",
    "            img = 255. - img\n",
    "        elif process_type == 2:\n",
    "            img = img.astype(float)\n",
    "            pass # do nothing\n",
    "        elif process_type == 3:\n",
    "            img = img.astype(float)\n",
    "            img = 255. - img\n",
    "        \n",
    "        return img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "dsb_dir = TRAIN_PATH\n",
    "test_dir = TEST_PATH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:25: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n"
     ]
    }
   ],
   "source": [
    "'''\n",
    "for i in range(2):\n",
    "    print(meta_train_val_groups[i][0].shape, meta_train_val_groups[i][1].shape)\n",
    "'''\n",
    "\n",
    "datasets = []\n",
    "for i in range(1):\n",
    "    \n",
    "    tr, val = meta_train_val_groups[i]\n",
    "\n",
    "    # Training dataset\n",
    "    dataset_train = DsbDataset()\n",
    "    dataset_train.load_dataset(tr[['ImageId', 'process_type']], augment=True)\n",
    "    dataset_train.prepare()\n",
    "\n",
    "    # Validation dataset\n",
    "    dataset_val = DsbDataset()\n",
    "    dataset_val.load_dataset(val[['ImageId', 'process_type']])\n",
    "    dataset_val.prepare()\n",
    "    \n",
    "    datasets.append((dataset_train, dataset_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at some images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading group= 0\n",
      "image shape= (696, 520, 3) mask shape= (696, 520, 23) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (360, 360, 3) mask shape= (360, 360, 26) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (1024, 1024, 3) mask shape= (1024, 1024, 57) bg type= white_black process type= 3\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7285: UserWarning: module 'imageio.plugins._tifffile' has no attribute 'decode_packbits'\n",
      "  Functionality might be degraded or be slow.\n",
      "\n",
      "  warnings.warn(\"%s%s\" % (e, warn))\n",
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7285: UserWarning: module 'imageio.plugins._tifffile' has no attribute 'decode_lzw'\n",
      "  Functionality might be degraded or be slow.\n",
      "\n",
      "  warnings.warn(\"%s%s\" % (e, warn))\n",
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7285: UserWarning: module 'imageio.plugins._tifffile' has no attribute 'unpack_ints'\n",
      "  Functionality might be degraded or be slow.\n",
      "\n",
      "  warnings.warn(\"%s%s\" % (e, warn))\n",
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7285: UserWarning: module 'imageio.plugins._tifffile' has no attribute 'reverse_bitorder'\n",
      "  Functionality might be degraded or be slow.\n",
      "\n",
      "  warnings.warn(\"%s%s\" % (e, warn))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (512, 512, 3) mask shape= (512, 512, 88) bg type= purple_purple process type= 1\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 320, 3) mask shape= (256, 320, 51) bg type= white_purple process type= 1\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 256, 3) mask shape= (256, 256, 8) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 320, 3) mask shape= (256, 320, 30) bg type= purple_purple process type= 1\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 256, 3) mask shape= (256, 256, 56) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 256, 3) mask shape= (256, 256, 32) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "image shape= (256, 256, 3) mask shape= (256, 256, 53) bg type= black_white process type= 2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(1):    \n",
    "    dataset_train, _ = datasets[i]\n",
    "    \n",
    "    #image_ids = dataset_train.image_ids\n",
    "    image_ids = np.random.choice(dataset_train.image_ids, 10)\n",
    "    print('Loading group=',i)\n",
    "    for image_id in image_ids:\n",
    "        image = dataset_train.load_image(image_id)\n",
    "\n",
    "        mask, class_ids = dataset_train.load_mask(image_id)\n",
    "        \n",
    "        print('image shape=', image.shape, 'mask shape=', mask.shape, \n",
    "              'bg type=', meta_train_val_groups[i][0].bg_type.values[image_id],\n",
    "              'process type=', meta_train_val_groups[i][0].process_type.values[image_id])\n",
    "        visualize.display_top_masks(image, mask, class_ids, dataset_train.class_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model Training\n",
    "\n",
    "OK! \n",
    "\n",
    "I think we are ready to do some training!\n",
    "We will instantiate the model and initialize it with COCO weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/base.py:198: retry (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use the retry module or similar alternatives.\n",
      "\n",
      "Starting at epoch 0. LR=0.001\n",
      "\n",
      "Checkpoint Path: /data/khyeh/logs/dsb20180410T0123/mask_rcnn_dsb_{epoch:04d}.h5\n",
      "Selecting layers to train\n",
      "conv1                  (Conv2D)\n",
      "bn_conv1               (BatchNorm)\n",
      "res2a_branch2a         (Conv2D)\n",
      "bn2a_branch2a          (BatchNorm)\n",
      "res2a_branch2b         (Conv2D)\n",
      "bn2a_branch2b          (BatchNorm)\n",
      "res2a_branch2c         (Conv2D)\n",
      "res2a_branch1          (Conv2D)\n",
      "bn2a_branch2c          (BatchNorm)\n",
      "bn2a_branch1           (BatchNorm)\n",
      "res2b_branch2a         (Conv2D)\n",
      "bn2b_branch2a          (BatchNorm)\n",
      "res2b_branch2b         (Conv2D)\n",
      "bn2b_branch2b          (BatchNorm)\n",
      "res2b_branch2c         (Conv2D)\n",
      "bn2b_branch2c          (BatchNorm)\n",
      "res2c_branch2a         (Conv2D)\n",
      "bn2c_branch2a          (BatchNorm)\n",
      "res2c_branch2b         (Conv2D)\n",
      "bn2c_branch2b          (BatchNorm)\n",
      "res2c_branch2c         (Conv2D)\n",
      "bn2c_branch2c          (BatchNorm)\n",
      "res3a_branch2a         (Conv2D)\n",
      "bn3a_branch2a          (BatchNorm)\n",
      "res3a_branch2b         (Conv2D)\n",
      "bn3a_branch2b          (BatchNorm)\n",
      "res3a_branch2c         (Conv2D)\n",
      "res3a_branch1          (Conv2D)\n",
      "bn3a_branch2c          (BatchNorm)\n",
      "bn3a_branch1           (BatchNorm)\n",
      "res3b_branch2a         (Conv2D)\n",
      "bn3b_branch2a          (BatchNorm)\n",
      "res3b_branch2b         (Conv2D)\n",
      "bn3b_branch2b          (BatchNorm)\n",
      "res3b_branch2c         (Conv2D)\n",
      "bn3b_branch2c          (BatchNorm)\n",
      "res3c_branch2a         (Conv2D)\n",
      "bn3c_branch2a          (BatchNorm)\n",
      "res3c_branch2b         (Conv2D)\n",
      "bn3c_branch2b          (BatchNorm)\n",
      "res3c_branch2c         (Conv2D)\n",
      "bn3c_branch2c          (BatchNorm)\n",
      "res3d_branch2a         (Conv2D)\n",
      "bn3d_branch2a          (BatchNorm)\n",
      "res3d_branch2b         (Conv2D)\n",
      "bn3d_branch2b          (BatchNorm)\n",
      "res3d_branch2c         (Conv2D)\n",
      "bn3d_branch2c          (BatchNorm)\n",
      "res4a_branch2a         (Conv2D)\n",
      "bn4a_branch2a          (BatchNorm)\n",
      "res4a_branch2b         (Conv2D)\n",
      "bn4a_branch2b          (BatchNorm)\n",
      "res4a_branch2c         (Conv2D)\n",
      "res4a_branch1          (Conv2D)\n",
      "bn4a_branch2c          (BatchNorm)\n",
      "bn4a_branch1           (BatchNorm)\n",
      "res4b_branch2a         (Conv2D)\n",
      "bn4b_branch2a          (BatchNorm)\n",
      "res4b_branch2b         (Conv2D)\n",
      "bn4b_branch2b          (BatchNorm)\n",
      "res4b_branch2c         (Conv2D)\n",
      "bn4b_branch2c          (BatchNorm)\n",
      "res4c_branch2a         (Conv2D)\n",
      "bn4c_branch2a          (BatchNorm)\n",
      "res4c_branch2b         (Conv2D)\n",
      "bn4c_branch2b          (BatchNorm)\n",
      "res4c_branch2c         (Conv2D)\n",
      "bn4c_branch2c          (BatchNorm)\n",
      "res4d_branch2a         (Conv2D)\n",
      "bn4d_branch2a          (BatchNorm)\n",
      "res4d_branch2b         (Conv2D)\n",
      "bn4d_branch2b          (BatchNorm)\n",
      "res4d_branch2c         (Conv2D)\n",
      "bn4d_branch2c          (BatchNorm)\n",
      "res4e_branch2a         (Conv2D)\n",
      "bn4e_branch2a          (BatchNorm)\n",
      "res4e_branch2b         (Conv2D)\n",
      "bn4e_branch2b          (BatchNorm)\n",
      "res4e_branch2c         (Conv2D)\n",
      "bn4e_branch2c          (BatchNorm)\n",
      "res4f_branch2a         (Conv2D)\n",
      "bn4f_branch2a          (BatchNorm)\n",
      "res4f_branch2b         (Conv2D)\n",
      "bn4f_branch2b          (BatchNorm)\n",
      "res4f_branch2c         (Conv2D)\n",
      "bn4f_branch2c          (BatchNorm)\n",
      "res4g_branch2a         (Conv2D)\n",
      "bn4g_branch2a          (BatchNorm)\n",
      "res4g_branch2b         (Conv2D)\n",
      "bn4g_branch2b          (BatchNorm)\n",
      "res4g_branch2c         (Conv2D)\n",
      "bn4g_branch2c          (BatchNorm)\n",
      "res4h_branch2a         (Conv2D)\n",
      "bn4h_branch2a          (BatchNorm)\n",
      "res4h_branch2b         (Conv2D)\n",
      "bn4h_branch2b          (BatchNorm)\n",
      "res4h_branch2c         (Conv2D)\n",
      "bn4h_branch2c          (BatchNorm)\n",
      "res4i_branch2a         (Conv2D)\n",
      "bn4i_branch2a          (BatchNorm)\n",
      "res4i_branch2b         (Conv2D)\n",
      "bn4i_branch2b          (BatchNorm)\n",
      "res4i_branch2c         (Conv2D)\n",
      "bn4i_branch2c          (BatchNorm)\n",
      "res4j_branch2a         (Conv2D)\n",
      "bn4j_branch2a          (BatchNorm)\n",
      "res4j_branch2b         (Conv2D)\n",
      "bn4j_branch2b          (BatchNorm)\n",
      "res4j_branch2c         (Conv2D)\n",
      "bn4j_branch2c          (BatchNorm)\n",
      "res4k_branch2a         (Conv2D)\n",
      "bn4k_branch2a          (BatchNorm)\n",
      "res4k_branch2b         (Conv2D)\n",
      "bn4k_branch2b          (BatchNorm)\n",
      "res4k_branch2c         (Conv2D)\n",
      "bn4k_branch2c          (BatchNorm)\n",
      "res4l_branch2a         (Conv2D)\n",
      "bn4l_branch2a          (BatchNorm)\n",
      "res4l_branch2b         (Conv2D)\n",
      "bn4l_branch2b          (BatchNorm)\n",
      "res4l_branch2c         (Conv2D)\n",
      "bn4l_branch2c          (BatchNorm)\n",
      "res4m_branch2a         (Conv2D)\n",
      "bn4m_branch2a          (BatchNorm)\n",
      "res4m_branch2b         (Conv2D)\n",
      "bn4m_branch2b          (BatchNorm)\n",
      "res4m_branch2c         (Conv2D)\n",
      "bn4m_branch2c          (BatchNorm)\n",
      "res4n_branch2a         (Conv2D)\n",
      "bn4n_branch2a          (BatchNorm)\n",
      "res4n_branch2b         (Conv2D)\n",
      "bn4n_branch2b          (BatchNorm)\n",
      "res4n_branch2c         (Conv2D)\n",
      "bn4n_branch2c          (BatchNorm)\n",
      "res4o_branch2a         (Conv2D)\n",
      "bn4o_branch2a          (BatchNorm)\n",
      "res4o_branch2b         (Conv2D)\n",
      "bn4o_branch2b          (BatchNorm)\n",
      "res4o_branch2c         (Conv2D)\n",
      "bn4o_branch2c          (BatchNorm)\n",
      "res4p_branch2a         (Conv2D)\n",
      "bn4p_branch2a          (BatchNorm)\n",
      "res4p_branch2b         (Conv2D)\n",
      "bn4p_branch2b          (BatchNorm)\n",
      "res4p_branch2c         (Conv2D)\n",
      "bn4p_branch2c          (BatchNorm)\n",
      "res4q_branch2a         (Conv2D)\n",
      "bn4q_branch2a          (BatchNorm)\n",
      "res4q_branch2b         (Conv2D)\n",
      "bn4q_branch2b          (BatchNorm)\n",
      "res4q_branch2c         (Conv2D)\n",
      "bn4q_branch2c          (BatchNorm)\n",
      "res4r_branch2a         (Conv2D)\n",
      "bn4r_branch2a          (BatchNorm)\n",
      "res4r_branch2b         (Conv2D)\n",
      "bn4r_branch2b          (BatchNorm)\n",
      "res4r_branch2c         (Conv2D)\n",
      "bn4r_branch2c          (BatchNorm)\n",
      "res4s_branch2a         (Conv2D)\n",
      "bn4s_branch2a          (BatchNorm)\n",
      "res4s_branch2b         (Conv2D)\n",
      "bn4s_branch2b          (BatchNorm)\n",
      "res4s_branch2c         (Conv2D)\n",
      "bn4s_branch2c          (BatchNorm)\n",
      "res4t_branch2a         (Conv2D)\n",
      "bn4t_branch2a          (BatchNorm)\n",
      "res4t_branch2b         (Conv2D)\n",
      "bn4t_branch2b          (BatchNorm)\n",
      "res4t_branch2c         (Conv2D)\n",
      "bn4t_branch2c          (BatchNorm)\n",
      "res4u_branch2a         (Conv2D)\n",
      "bn4u_branch2a          (BatchNorm)\n",
      "res4u_branch2b         (Conv2D)\n",
      "bn4u_branch2b          (BatchNorm)\n",
      "res4u_branch2c         (Conv2D)\n",
      "bn4u_branch2c          (BatchNorm)\n",
      "res4v_branch2a         (Conv2D)\n",
      "bn4v_branch2a          (BatchNorm)\n",
      "res4v_branch2b         (Conv2D)\n",
      "bn4v_branch2b          (BatchNorm)\n",
      "res4v_branch2c         (Conv2D)\n",
      "bn4v_branch2c          (BatchNorm)\n",
      "res4w_branch2a         (Conv2D)\n",
      "bn4w_branch2a          (BatchNorm)\n",
      "res4w_branch2b         (Conv2D)\n",
      "bn4w_branch2b          (BatchNorm)\n",
      "res4w_branch2c         (Conv2D)\n",
      "bn4w_branch2c          (BatchNorm)\n",
      "res5a_branch2a         (Conv2D)\n",
      "bn5a_branch2a          (BatchNorm)\n",
      "res5a_branch2b         (Conv2D)\n",
      "bn5a_branch2b          (BatchNorm)\n",
      "res5a_branch2c         (Conv2D)\n",
      "res5a_branch1          (Conv2D)\n",
      "bn5a_branch2c          (BatchNorm)\n",
      "bn5a_branch1           (BatchNorm)\n",
      "res5b_branch2a         (Conv2D)\n",
      "bn5b_branch2a          (BatchNorm)\n",
      "res5b_branch2b         (Conv2D)\n",
      "bn5b_branch2b          (BatchNorm)\n",
      "res5b_branch2c         (Conv2D)\n",
      "bn5b_branch2c          (BatchNorm)\n",
      "res5c_branch2a         (Conv2D)\n",
      "bn5c_branch2a          (BatchNorm)\n",
      "res5c_branch2b         (Conv2D)\n",
      "bn5c_branch2b          (BatchNorm)\n",
      "res5c_branch2c         (Conv2D)\n",
      "bn5c_branch2c          (BatchNorm)\n",
      "fpn_c5p5               (Conv2D)\n",
      "fpn_c4p4               (Conv2D)\n",
      "fpn_c3p3               (Conv2D)\n",
      "fpn_c2p2               (Conv2D)\n",
      "fpn_p5                 (Conv2D)\n",
      "fpn_p2                 (Conv2D)\n",
      "fpn_p3                 (Conv2D)\n",
      "fpn_p4                 (Conv2D)\n",
      "In model:  rpn_model\n",
      "    rpn_conv_shared        (Conv2D)\n",
      "    rpn_class_raw          (Conv2D)\n",
      "    rpn_bbox_pred          (Conv2D)\n",
      "mrcnn_mask_conv1       (TimeDistributed)\n",
      "mrcnn_mask_bn1         (TimeDistributed)\n",
      "mrcnn_mask_conv2       (TimeDistributed)\n",
      "mrcnn_mask_bn2         (TimeDistributed)\n",
      "mrcnn_class_conv1      (TimeDistributed)\n",
      "mrcnn_class_bn1        (TimeDistributed)\n",
      "mrcnn_mask_conv3       (TimeDistributed)\n",
      "mrcnn_mask_bn3         (TimeDistributed)\n",
      "mrcnn_class_conv2      (TimeDistributed)\n",
      "mrcnn_class_bn2        (TimeDistributed)\n",
      "mrcnn_mask_conv4       (TimeDistributed)\n",
      "mrcnn_mask_bn4         (TimeDistributed)\n",
      "mrcnn_bbox_fc          (TimeDistributed)\n",
      "mrcnn_mask_deconv      (TimeDistributed)\n",
      "mrcnn_class_logits     (TimeDistributed)\n",
      "mrcnn_mask             (TimeDistributed)\n",
      "WARNING:tensorflow:From Mask_RCNN/model.py:2157: calling reduce_mean (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "keep_dims is deprecated, use keepdims instead\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/gradients_impl.py:100: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
      "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n",
      "/usr/local/lib/python3.5/dist-packages/keras/engine/training.py:2087: UserWarning: Using a generator with `use_multiprocessing=True` and multiple workers may duplicate your data. Please consider using the`keras.utils.Sequence class.\n",
      "  UserWarning('Using a generator with `use_multiprocessing=True`'\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/scipy/ndimage/interpolation.py:616: UserWarning: From scipy 0.13.0, the output shape of zoom() is calculated with round() instead of int() - for these inputs the size of the returned array has changed.\n",
      "  \"the returned array has changed.\", UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch 00001: LearningRateScheduler reducing learning rate to 0.001.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/scipy/ndimage/interpolation.py:616: UserWarning: From scipy 0.13.0, the output shape of zoom() is calculated with round() instead of int() - for these inputs the size of the returned array has changed.\n",
      "  \"the returned array has changed.\", UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "116/343 [=========>....................] - ETA: 8:58 - loss: 1.8627 - rpn_class_loss: 0.2236 - rpn_bbox_loss: 0.8088 - mrcnn_class_loss: 0.0998 - mrcnn_bbox_loss: 0.4054 - mrcnn_mask_loss: 0.3252"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7472: UserWarning: unexpected end of lzw stream (code 0)\n",
      "  warnings.warn(\"unexpected end of lzw stream (code %i)\" % code)\n",
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:3284: UserWarning: invalid tile data\n",
      "  warnings.warn(\"invalid tile data\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "203/343 [================>.............] - ETA: 5:25 - loss: 1.6398 - rpn_class_loss: 0.1704 - rpn_bbox_loss: 0.7529 - mrcnn_class_loss: 0.0983 - mrcnn_bbox_loss: 0.3255 - mrcnn_mask_loss: 0.2927"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:7472: UserWarning: unexpected end of lzw stream (code 0)\n",
      "  warnings.warn(\"unexpected end of lzw stream (code %i)\" % code)\n",
      "/usr/local/lib/python3.5/dist-packages/imageio/plugins/_tifffile.py:3284: UserWarning: invalid tile data\n",
      "  warnings.warn(\"invalid tile data\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "343/343 [==============================] - 791s 2s/step - loss: 1.4513 - rpn_class_loss: 0.1284 - rpn_bbox_loss: 0.6621 - mrcnn_class_loss: 0.1110 - mrcnn_bbox_loss: 0.2758 - mrcnn_mask_loss: 0.2741 - val_loss: 0.8218 - val_rpn_class_loss: 0.0315 - val_rpn_bbox_loss: 0.3929 - val_mrcnn_class_loss: 0.1101 - val_mrcnn_bbox_loss: 0.1248 - val_mrcnn_mask_loss: 0.1626\n",
      "Epoch 2/100\n",
      "\n",
      "Epoch 00002: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.9561 - rpn_class_loss: 0.0478 - rpn_bbox_loss: 0.3714 - mrcnn_class_loss: 0.1177 - mrcnn_bbox_loss: 0.1829 - mrcnn_mask_loss: 0.2363 - val_loss: 0.6356 - val_rpn_class_loss: 0.0157 - val_rpn_bbox_loss: 0.2721 - val_mrcnn_class_loss: 0.0913 - val_mrcnn_bbox_loss: 0.1031 - val_mrcnn_mask_loss: 0.1534\n",
      "Epoch 3/100\n",
      "\n",
      "Epoch 00003: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 775s 2s/step - loss: 0.8707 - rpn_class_loss: 0.0413 - rpn_bbox_loss: 0.3128 - mrcnn_class_loss: 0.1131 - mrcnn_bbox_loss: 0.1702 - mrcnn_mask_loss: 0.2333 - val_loss: 0.5592 - val_rpn_class_loss: 0.0164 - val_rpn_bbox_loss: 0.2055 - val_mrcnn_class_loss: 0.0734 - val_mrcnn_bbox_loss: 0.1033 - val_mrcnn_mask_loss: 0.1607\n",
      "Epoch 4/100\n",
      "\n",
      "Epoch 00004: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.8204 - rpn_class_loss: 0.0358 - rpn_bbox_loss: 0.2825 - mrcnn_class_loss: 0.1086 - mrcnn_bbox_loss: 0.1639 - mrcnn_mask_loss: 0.2297 - val_loss: 0.5049 - val_rpn_class_loss: 0.0103 - val_rpn_bbox_loss: 0.1618 - val_mrcnn_class_loss: 0.0817 - val_mrcnn_bbox_loss: 0.0974 - val_mrcnn_mask_loss: 0.1536\n",
      "Epoch 5/100\n",
      "\n",
      "Epoch 00005: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 775s 2s/step - loss: 0.8001 - rpn_class_loss: 0.0336 - rpn_bbox_loss: 0.2636 - mrcnn_class_loss: 0.1078 - mrcnn_bbox_loss: 0.1628 - mrcnn_mask_loss: 0.2324 - val_loss: 0.5012 - val_rpn_class_loss: 0.0091 - val_rpn_bbox_loss: 0.1595 - val_mrcnn_class_loss: 0.0725 - val_mrcnn_bbox_loss: 0.1019 - val_mrcnn_mask_loss: 0.1581\n",
      "Epoch 6/100\n",
      "\n",
      "Epoch 00006: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 775s 2s/step - loss: 0.7646 - rpn_class_loss: 0.0316 - rpn_bbox_loss: 0.2471 - mrcnn_class_loss: 0.1072 - mrcnn_bbox_loss: 0.1522 - mrcnn_mask_loss: 0.2265 - val_loss: 0.4387 - val_rpn_class_loss: 0.0086 - val_rpn_bbox_loss: 0.1367 - val_mrcnn_class_loss: 0.0584 - val_mrcnn_bbox_loss: 0.0800 - val_mrcnn_mask_loss: 0.1549\n",
      "Epoch 7/100\n",
      "\n",
      "Epoch 00007: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.7398 - rpn_class_loss: 0.0300 - rpn_bbox_loss: 0.2404 - mrcnn_class_loss: 0.0978 - mrcnn_bbox_loss: 0.1470 - mrcnn_mask_loss: 0.2247 - val_loss: 0.4524 - val_rpn_class_loss: 0.0104 - val_rpn_bbox_loss: 0.1346 - val_mrcnn_class_loss: 0.0681 - val_mrcnn_bbox_loss: 0.0793 - val_mrcnn_mask_loss: 0.1602\n",
      "Epoch 8/100\n",
      "\n",
      "Epoch 00008: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.7342 - rpn_class_loss: 0.0296 - rpn_bbox_loss: 0.2318 - mrcnn_class_loss: 0.1018 - mrcnn_bbox_loss: 0.1469 - mrcnn_mask_loss: 0.2241 - val_loss: 0.4424 - val_rpn_class_loss: 0.0073 - val_rpn_bbox_loss: 0.1135 - val_mrcnn_class_loss: 0.0928 - val_mrcnn_bbox_loss: 0.0784 - val_mrcnn_mask_loss: 0.1503\n",
      "Epoch 9/100\n",
      "\n",
      "Epoch 00009: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.7252 - rpn_class_loss: 0.0281 - rpn_bbox_loss: 0.2222 - mrcnn_class_loss: 0.1009 - mrcnn_bbox_loss: 0.1473 - mrcnn_mask_loss: 0.2266 - val_loss: 0.4566 - val_rpn_class_loss: 0.0083 - val_rpn_bbox_loss: 0.1139 - val_mrcnn_class_loss: 0.0743 - val_mrcnn_bbox_loss: 0.0929 - val_mrcnn_mask_loss: 0.1673\n",
      "Epoch 10/100\n",
      "\n",
      "Epoch 00010: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.6861 - rpn_class_loss: 0.0258 - rpn_bbox_loss: 0.2117 - mrcnn_class_loss: 0.0925 - mrcnn_bbox_loss: 0.1378 - mrcnn_mask_loss: 0.2184 - val_loss: 0.4322 - val_rpn_class_loss: 0.0111 - val_rpn_bbox_loss: 0.1032 - val_mrcnn_class_loss: 0.0646 - val_mrcnn_bbox_loss: 0.0936 - val_mrcnn_mask_loss: 0.1597\n",
      "Epoch 11/100\n",
      "\n",
      "Epoch 00011: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.6935 - rpn_class_loss: 0.0261 - rpn_bbox_loss: 0.2099 - mrcnn_class_loss: 0.0954 - mrcnn_bbox_loss: 0.1397 - mrcnn_mask_loss: 0.2223 - val_loss: 0.4369 - val_rpn_class_loss: 0.0117 - val_rpn_bbox_loss: 0.1174 - val_mrcnn_class_loss: 0.0633 - val_mrcnn_bbox_loss: 0.0791 - val_mrcnn_mask_loss: 0.1654\n",
      "Epoch 12/100\n",
      "\n",
      "Epoch 00012: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.6892 - rpn_class_loss: 0.0251 - rpn_bbox_loss: 0.2082 - mrcnn_class_loss: 0.0919 - mrcnn_bbox_loss: 0.1413 - mrcnn_mask_loss: 0.2227 - val_loss: 0.4120 - val_rpn_class_loss: 0.0086 - val_rpn_bbox_loss: 0.1124 - val_mrcnn_class_loss: 0.0739 - val_mrcnn_bbox_loss: 0.0688 - val_mrcnn_mask_loss: 0.1483\n",
      "Epoch 13/100\n",
      "\n",
      "Epoch 00013: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6515 - rpn_class_loss: 0.0230 - rpn_bbox_loss: 0.1990 - mrcnn_class_loss: 0.0863 - mrcnn_bbox_loss: 0.1298 - mrcnn_mask_loss: 0.2133 - val_loss: 0.4153 - val_rpn_class_loss: 0.0104 - val_rpn_bbox_loss: 0.1167 - val_mrcnn_class_loss: 0.0559 - val_mrcnn_bbox_loss: 0.0769 - val_mrcnn_mask_loss: 0.1554\n",
      "Epoch 14/100\n",
      "\n",
      "Epoch 00014: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.7037 - rpn_class_loss: 0.0250 - rpn_bbox_loss: 0.2131 - mrcnn_class_loss: 0.0955 - mrcnn_bbox_loss: 0.1430 - mrcnn_mask_loss: 0.2271 - val_loss: 0.4104 - val_rpn_class_loss: 0.0098 - val_rpn_bbox_loss: 0.1052 - val_mrcnn_class_loss: 0.0575 - val_mrcnn_bbox_loss: 0.0763 - val_mrcnn_mask_loss: 0.1615\n",
      "Epoch 15/100\n",
      "\n",
      "Epoch 00015: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.6698 - rpn_class_loss: 0.0240 - rpn_bbox_loss: 0.2003 - mrcnn_class_loss: 0.0898 - mrcnn_bbox_loss: 0.1347 - mrcnn_mask_loss: 0.2211 - val_loss: 0.4043 - val_rpn_class_loss: 0.0111 - val_rpn_bbox_loss: 0.1028 - val_mrcnn_class_loss: 0.0513 - val_mrcnn_bbox_loss: 0.0818 - val_mrcnn_mask_loss: 0.1574\n",
      "Epoch 16/100\n",
      "\n",
      "Epoch 00016: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.6508 - rpn_class_loss: 0.0225 - rpn_bbox_loss: 0.1955 - mrcnn_class_loss: 0.0845 - mrcnn_bbox_loss: 0.1315 - mrcnn_mask_loss: 0.2169 - val_loss: 0.4460 - val_rpn_class_loss: 0.0091 - val_rpn_bbox_loss: 0.1298 - val_mrcnn_class_loss: 0.0565 - val_mrcnn_bbox_loss: 0.0918 - val_mrcnn_mask_loss: 0.1588\n",
      "Epoch 17/100\n",
      "\n",
      "Epoch 00017: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6559 - rpn_class_loss: 0.0223 - rpn_bbox_loss: 0.1955 - mrcnn_class_loss: 0.0858 - mrcnn_bbox_loss: 0.1328 - mrcnn_mask_loss: 0.2195 - val_loss: 0.4532 - val_rpn_class_loss: 0.0183 - val_rpn_bbox_loss: 0.1268 - val_mrcnn_class_loss: 0.0654 - val_mrcnn_bbox_loss: 0.0803 - val_mrcnn_mask_loss: 0.1624\n",
      "Epoch 18/100\n",
      "\n",
      "Epoch 00018: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.6496 - rpn_class_loss: 0.0228 - rpn_bbox_loss: 0.1920 - mrcnn_class_loss: 0.0875 - mrcnn_bbox_loss: 0.1298 - mrcnn_mask_loss: 0.2175 - val_loss: 0.4230 - val_rpn_class_loss: 0.0115 - val_rpn_bbox_loss: 0.1152 - val_mrcnn_class_loss: 0.0461 - val_mrcnn_bbox_loss: 0.0878 - val_mrcnn_mask_loss: 0.1624\n",
      "Epoch 19/100\n",
      "\n",
      "Epoch 00019: LearningRateScheduler reducing learning rate to 0.001.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "343/343 [==============================] - 775s 2s/step - loss: 0.6352 - rpn_class_loss: 0.0220 - rpn_bbox_loss: 0.1839 - mrcnn_class_loss: 0.0835 - mrcnn_bbox_loss: 0.1281 - mrcnn_mask_loss: 0.2177 - val_loss: 0.4005 - val_rpn_class_loss: 0.0115 - val_rpn_bbox_loss: 0.0907 - val_mrcnn_class_loss: 0.0628 - val_mrcnn_bbox_loss: 0.0749 - val_mrcnn_mask_loss: 0.1606\n",
      "Epoch 20/100\n",
      "\n",
      "Epoch 00020: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6202 - rpn_class_loss: 0.0206 - rpn_bbox_loss: 0.1824 - mrcnn_class_loss: 0.0840 - mrcnn_bbox_loss: 0.1226 - mrcnn_mask_loss: 0.2106 - val_loss: 0.4345 - val_rpn_class_loss: 0.0132 - val_rpn_bbox_loss: 0.1136 - val_mrcnn_class_loss: 0.0567 - val_mrcnn_bbox_loss: 0.0856 - val_mrcnn_mask_loss: 0.1653\n",
      "Epoch 21/100\n",
      "\n",
      "Epoch 00021: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.6414 - rpn_class_loss: 0.0221 - rpn_bbox_loss: 0.1869 - mrcnn_class_loss: 0.0862 - mrcnn_bbox_loss: 0.1286 - mrcnn_mask_loss: 0.2176 - val_loss: 0.4211 - val_rpn_class_loss: 0.0141 - val_rpn_bbox_loss: 0.1012 - val_mrcnn_class_loss: 0.0636 - val_mrcnn_bbox_loss: 0.0812 - val_mrcnn_mask_loss: 0.1609\n",
      "Epoch 22/100\n",
      "\n",
      "Epoch 00022: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 781s 2s/step - loss: 0.6308 - rpn_class_loss: 0.0214 - rpn_bbox_loss: 0.1856 - mrcnn_class_loss: 0.0830 - mrcnn_bbox_loss: 0.1262 - mrcnn_mask_loss: 0.2147 - val_loss: 0.3280 - val_rpn_class_loss: 0.0062 - val_rpn_bbox_loss: 0.0704 - val_mrcnn_class_loss: 0.0537 - val_mrcnn_bbox_loss: 0.0556 - val_mrcnn_mask_loss: 0.1422\n",
      "Epoch 23/100\n",
      "\n",
      "Epoch 00023: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.6151 - rpn_class_loss: 0.0192 - rpn_bbox_loss: 0.1768 - mrcnn_class_loss: 0.0812 - mrcnn_bbox_loss: 0.1241 - mrcnn_mask_loss: 0.2137 - val_loss: 0.3848 - val_rpn_class_loss: 0.0171 - val_rpn_bbox_loss: 0.1009 - val_mrcnn_class_loss: 0.0530 - val_mrcnn_bbox_loss: 0.0634 - val_mrcnn_mask_loss: 0.1504\n",
      "Epoch 24/100\n",
      "\n",
      "Epoch 00024: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 782s 2s/step - loss: 0.6292 - rpn_class_loss: 0.0211 - rpn_bbox_loss: 0.1829 - mrcnn_class_loss: 0.0831 - mrcnn_bbox_loss: 0.1264 - mrcnn_mask_loss: 0.2158 - val_loss: 0.3776 - val_rpn_class_loss: 0.0111 - val_rpn_bbox_loss: 0.0850 - val_mrcnn_class_loss: 0.0636 - val_mrcnn_bbox_loss: 0.0653 - val_mrcnn_mask_loss: 0.1526\n",
      "Epoch 25/100\n",
      "\n",
      "Epoch 00025: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.6070 - rpn_class_loss: 0.0199 - rpn_bbox_loss: 0.1729 - mrcnn_class_loss: 0.0780 - mrcnn_bbox_loss: 0.1230 - mrcnn_mask_loss: 0.2132 - val_loss: 0.4192 - val_rpn_class_loss: 0.0152 - val_rpn_bbox_loss: 0.1020 - val_mrcnn_class_loss: 0.0606 - val_mrcnn_bbox_loss: 0.0822 - val_mrcnn_mask_loss: 0.1592\n",
      "Epoch 26/100\n",
      "\n",
      "Epoch 00026: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.6392 - rpn_class_loss: 0.0202 - rpn_bbox_loss: 0.1841 - mrcnn_class_loss: 0.0820 - mrcnn_bbox_loss: 0.1301 - mrcnn_mask_loss: 0.2227 - val_loss: 0.4152 - val_rpn_class_loss: 0.0120 - val_rpn_bbox_loss: 0.1029 - val_mrcnn_class_loss: 0.0608 - val_mrcnn_bbox_loss: 0.0772 - val_mrcnn_mask_loss: 0.1623\n",
      "Epoch 27/100\n",
      "\n",
      "Epoch 00027: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.6036 - rpn_class_loss: 0.0200 - rpn_bbox_loss: 0.1712 - mrcnn_class_loss: 0.0791 - mrcnn_bbox_loss: 0.1206 - mrcnn_mask_loss: 0.2127 - val_loss: 0.3535 - val_rpn_class_loss: 0.0093 - val_rpn_bbox_loss: 0.0849 - val_mrcnn_class_loss: 0.0432 - val_mrcnn_bbox_loss: 0.0671 - val_mrcnn_mask_loss: 0.1490\n",
      "Epoch 28/100\n",
      "\n",
      "Epoch 00028: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6146 - rpn_class_loss: 0.0185 - rpn_bbox_loss: 0.1738 - mrcnn_class_loss: 0.0810 - mrcnn_bbox_loss: 0.1245 - mrcnn_mask_loss: 0.2168 - val_loss: 0.3965 - val_rpn_class_loss: 0.0167 - val_rpn_bbox_loss: 0.1133 - val_mrcnn_class_loss: 0.0383 - val_mrcnn_bbox_loss: 0.0704 - val_mrcnn_mask_loss: 0.1579\n",
      "Epoch 29/100\n",
      "\n",
      "Epoch 00029: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6036 - rpn_class_loss: 0.0187 - rpn_bbox_loss: 0.1753 - mrcnn_class_loss: 0.0734 - mrcnn_bbox_loss: 0.1221 - mrcnn_mask_loss: 0.2141 - val_loss: 0.3876 - val_rpn_class_loss: 0.0139 - val_rpn_bbox_loss: 0.1064 - val_mrcnn_class_loss: 0.0457 - val_mrcnn_bbox_loss: 0.0666 - val_mrcnn_mask_loss: 0.1550\n",
      "Epoch 30/100\n",
      "\n",
      "Epoch 00030: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.6063 - rpn_class_loss: 0.0200 - rpn_bbox_loss: 0.1674 - mrcnn_class_loss: 0.0836 - mrcnn_bbox_loss: 0.1215 - mrcnn_mask_loss: 0.2138 - val_loss: 0.3849 - val_rpn_class_loss: 0.0134 - val_rpn_bbox_loss: 0.1009 - val_mrcnn_class_loss: 0.0408 - val_mrcnn_bbox_loss: 0.0679 - val_mrcnn_mask_loss: 0.1618\n",
      "Epoch 31/100\n",
      "\n",
      "Epoch 00031: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 775s 2s/step - loss: 0.5875 - rpn_class_loss: 0.0181 - rpn_bbox_loss: 0.1672 - mrcnn_class_loss: 0.0743 - mrcnn_bbox_loss: 0.1177 - mrcnn_mask_loss: 0.2101 - val_loss: 0.3808 - val_rpn_class_loss: 0.0111 - val_rpn_bbox_loss: 0.1050 - val_mrcnn_class_loss: 0.0439 - val_mrcnn_bbox_loss: 0.0701 - val_mrcnn_mask_loss: 0.1507\n",
      "Epoch 32/100\n",
      "\n",
      "Epoch 00032: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.6040 - rpn_class_loss: 0.0190 - rpn_bbox_loss: 0.1680 - mrcnn_class_loss: 0.0777 - mrcnn_bbox_loss: 0.1222 - mrcnn_mask_loss: 0.2170 - val_loss: 0.3730 - val_rpn_class_loss: 0.0105 - val_rpn_bbox_loss: 0.0943 - val_mrcnn_class_loss: 0.0324 - val_mrcnn_bbox_loss: 0.0679 - val_mrcnn_mask_loss: 0.1678\n",
      "Epoch 33/100\n",
      "\n",
      "Epoch 00033: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 783s 2s/step - loss: 0.6069 - rpn_class_loss: 0.0183 - rpn_bbox_loss: 0.1689 - mrcnn_class_loss: 0.0784 - mrcnn_bbox_loss: 0.1225 - mrcnn_mask_loss: 0.2188 - val_loss: 0.4354 - val_rpn_class_loss: 0.0115 - val_rpn_bbox_loss: 0.1164 - val_mrcnn_class_loss: 0.0393 - val_mrcnn_bbox_loss: 0.0808 - val_mrcnn_mask_loss: 0.1874\n",
      "Epoch 34/100\n",
      "\n",
      "Epoch 00034: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 783s 2s/step - loss: 0.6045 - rpn_class_loss: 0.0179 - rpn_bbox_loss: 0.1722 - mrcnn_class_loss: 0.0736 - mrcnn_bbox_loss: 0.1245 - mrcnn_mask_loss: 0.2164 - val_loss: 0.3521 - val_rpn_class_loss: 0.0103 - val_rpn_bbox_loss: 0.0911 - val_mrcnn_class_loss: 0.0247 - val_mrcnn_bbox_loss: 0.0682 - val_mrcnn_mask_loss: 0.1578\n",
      "Epoch 35/100\n",
      "\n",
      "Epoch 00035: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 784s 2s/step - loss: 0.6095 - rpn_class_loss: 0.0191 - rpn_bbox_loss: 0.1704 - mrcnn_class_loss: 0.0760 - mrcnn_bbox_loss: 0.1242 - mrcnn_mask_loss: 0.2198 - val_loss: 0.3987 - val_rpn_class_loss: 0.0124 - val_rpn_bbox_loss: 0.1018 - val_mrcnn_class_loss: 0.0410 - val_mrcnn_bbox_loss: 0.0796 - val_mrcnn_mask_loss: 0.1638\n",
      "Epoch 36/100\n",
      "\n",
      "Epoch 00036: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5688 - rpn_class_loss: 0.0176 - rpn_bbox_loss: 0.1539 - mrcnn_class_loss: 0.0703 - mrcnn_bbox_loss: 0.1156 - mrcnn_mask_loss: 0.2115 - val_loss: 0.4442 - val_rpn_class_loss: 0.0149 - val_rpn_bbox_loss: 0.1143 - val_mrcnn_class_loss: 0.0260 - val_mrcnn_bbox_loss: 0.1035 - val_mrcnn_mask_loss: 0.1855\n",
      "Epoch 37/100\n",
      "\n",
      "Epoch 00037: LearningRateScheduler reducing learning rate to 0.001.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "343/343 [==============================] - 776s 2s/step - loss: 0.5753 - rpn_class_loss: 0.0177 - rpn_bbox_loss: 0.1605 - mrcnn_class_loss: 0.0690 - mrcnn_bbox_loss: 0.1155 - mrcnn_mask_loss: 0.2127 - val_loss: 0.4130 - val_rpn_class_loss: 0.0208 - val_rpn_bbox_loss: 0.1240 - val_mrcnn_class_loss: 0.0327 - val_mrcnn_bbox_loss: 0.0672 - val_mrcnn_mask_loss: 0.1683\n",
      "Epoch 38/100\n",
      "\n",
      "Epoch 00038: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.5835 - rpn_class_loss: 0.0176 - rpn_bbox_loss: 0.1582 - mrcnn_class_loss: 0.0731 - mrcnn_bbox_loss: 0.1196 - mrcnn_mask_loss: 0.2150 - val_loss: 0.4133 - val_rpn_class_loss: 0.0191 - val_rpn_bbox_loss: 0.1275 - val_mrcnn_class_loss: 0.0256 - val_mrcnn_bbox_loss: 0.0670 - val_mrcnn_mask_loss: 0.1741\n",
      "Epoch 39/100\n",
      "\n",
      "Epoch 00039: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5778 - rpn_class_loss: 0.0166 - rpn_bbox_loss: 0.1584 - mrcnn_class_loss: 0.0701 - mrcnn_bbox_loss: 0.1167 - mrcnn_mask_loss: 0.2160 - val_loss: 0.4477 - val_rpn_class_loss: 0.0187 - val_rpn_bbox_loss: 0.1569 - val_mrcnn_class_loss: 0.0412 - val_mrcnn_bbox_loss: 0.0639 - val_mrcnn_mask_loss: 0.1670\n",
      "Epoch 40/100\n",
      "\n",
      "Epoch 00040: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.5750 - rpn_class_loss: 0.0175 - rpn_bbox_loss: 0.1570 - mrcnn_class_loss: 0.0702 - mrcnn_bbox_loss: 0.1165 - mrcnn_mask_loss: 0.2137 - val_loss: 0.4361 - val_rpn_class_loss: 0.0092 - val_rpn_bbox_loss: 0.0965 - val_mrcnn_class_loss: 0.0669 - val_mrcnn_bbox_loss: 0.0952 - val_mrcnn_mask_loss: 0.1683\n",
      "Epoch 41/100\n",
      "\n",
      "Epoch 00041: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5776 - rpn_class_loss: 0.0168 - rpn_bbox_loss: 0.1601 - mrcnn_class_loss: 0.0692 - mrcnn_bbox_loss: 0.1160 - mrcnn_mask_loss: 0.2154 - val_loss: 0.3917 - val_rpn_class_loss: 0.0092 - val_rpn_bbox_loss: 0.0856 - val_mrcnn_class_loss: 0.0459 - val_mrcnn_bbox_loss: 0.0811 - val_mrcnn_mask_loss: 0.1699\n",
      "Epoch 42/100\n",
      "\n",
      "Epoch 00042: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.5784 - rpn_class_loss: 0.0178 - rpn_bbox_loss: 0.1540 - mrcnn_class_loss: 0.0732 - mrcnn_bbox_loss: 0.1186 - mrcnn_mask_loss: 0.2148 - val_loss: 0.4315 - val_rpn_class_loss: 0.0171 - val_rpn_bbox_loss: 0.1401 - val_mrcnn_class_loss: 0.0180 - val_mrcnn_bbox_loss: 0.0891 - val_mrcnn_mask_loss: 0.1673\n",
      "Epoch 43/100\n",
      "\n",
      "Epoch 00043: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5480 - rpn_class_loss: 0.0164 - rpn_bbox_loss: 0.1473 - mrcnn_class_loss: 0.0675 - mrcnn_bbox_loss: 0.1094 - mrcnn_mask_loss: 0.2074 - val_loss: 0.4437 - val_rpn_class_loss: 0.0122 - val_rpn_bbox_loss: 0.1168 - val_mrcnn_class_loss: 0.0327 - val_mrcnn_bbox_loss: 0.1004 - val_mrcnn_mask_loss: 0.1816\n",
      "Epoch 44/100\n",
      "\n",
      "Epoch 00044: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5791 - rpn_class_loss: 0.0175 - rpn_bbox_loss: 0.1578 - mrcnn_class_loss: 0.0703 - mrcnn_bbox_loss: 0.1169 - mrcnn_mask_loss: 0.2166 - val_loss: 0.4632 - val_rpn_class_loss: 0.0125 - val_rpn_bbox_loss: 0.1141 - val_mrcnn_class_loss: 0.0460 - val_mrcnn_bbox_loss: 0.1081 - val_mrcnn_mask_loss: 0.1826\n",
      "Epoch 45/100\n",
      "\n",
      "Epoch 00045: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.5677 - rpn_class_loss: 0.0175 - rpn_bbox_loss: 0.1531 - mrcnn_class_loss: 0.0692 - mrcnn_bbox_loss: 0.1144 - mrcnn_mask_loss: 0.2134 - val_loss: 0.4932 - val_rpn_class_loss: 0.0219 - val_rpn_bbox_loss: 0.1612 - val_mrcnn_class_loss: 0.0336 - val_mrcnn_bbox_loss: 0.0970 - val_mrcnn_mask_loss: 0.1795\n",
      "Epoch 46/100\n",
      "\n",
      "Epoch 00046: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5484 - rpn_class_loss: 0.0150 - rpn_bbox_loss: 0.1480 - mrcnn_class_loss: 0.0627 - mrcnn_bbox_loss: 0.1116 - mrcnn_mask_loss: 0.2112 - val_loss: 0.4766 - val_rpn_class_loss: 0.0213 - val_rpn_bbox_loss: 0.1587 - val_mrcnn_class_loss: 0.0422 - val_mrcnn_bbox_loss: 0.0837 - val_mrcnn_mask_loss: 0.1707\n",
      "Epoch 47/100\n",
      "\n",
      "Epoch 00047: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 782s 2s/step - loss: 0.5889 - rpn_class_loss: 0.0172 - rpn_bbox_loss: 0.1600 - mrcnn_class_loss: 0.0697 - mrcnn_bbox_loss: 0.1197 - mrcnn_mask_loss: 0.2223 - val_loss: 0.4318 - val_rpn_class_loss: 0.0200 - val_rpn_bbox_loss: 0.1327 - val_mrcnn_class_loss: 0.0231 - val_mrcnn_bbox_loss: 0.0761 - val_mrcnn_mask_loss: 0.1799\n",
      "Epoch 48/100\n",
      "\n",
      "Epoch 00048: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 782s 2s/step - loss: 0.5211 - rpn_class_loss: 0.0141 - rpn_bbox_loss: 0.1355 - mrcnn_class_loss: 0.0633 - mrcnn_bbox_loss: 0.1042 - mrcnn_mask_loss: 0.2039 - val_loss: 0.4437 - val_rpn_class_loss: 0.0182 - val_rpn_bbox_loss: 0.1450 - val_mrcnn_class_loss: 0.0339 - val_mrcnn_bbox_loss: 0.0787 - val_mrcnn_mask_loss: 0.1679\n",
      "Epoch 49/100\n",
      "\n",
      "Epoch 00049: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 782s 2s/step - loss: 0.5609 - rpn_class_loss: 0.0160 - rpn_bbox_loss: 0.1516 - mrcnn_class_loss: 0.0654 - mrcnn_bbox_loss: 0.1128 - mrcnn_mask_loss: 0.2149 - val_loss: 0.4490 - val_rpn_class_loss: 0.0135 - val_rpn_bbox_loss: 0.1274 - val_mrcnn_class_loss: 0.0243 - val_mrcnn_bbox_loss: 0.1003 - val_mrcnn_mask_loss: 0.1834\n",
      "Epoch 50/100\n",
      "\n",
      "Epoch 00050: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5427 - rpn_class_loss: 0.0166 - rpn_bbox_loss: 0.1419 - mrcnn_class_loss: 0.0655 - mrcnn_bbox_loss: 0.1090 - mrcnn_mask_loss: 0.2097 - val_loss: 0.4948 - val_rpn_class_loss: 0.0190 - val_rpn_bbox_loss: 0.1491 - val_mrcnn_class_loss: 0.0327 - val_mrcnn_bbox_loss: 0.0965 - val_mrcnn_mask_loss: 0.1975\n",
      "Epoch 51/100\n",
      "\n",
      "Epoch 00051: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5511 - rpn_class_loss: 0.0158 - rpn_bbox_loss: 0.1492 - mrcnn_class_loss: 0.0627 - mrcnn_bbox_loss: 0.1117 - mrcnn_mask_loss: 0.2117 - val_loss: 0.4303 - val_rpn_class_loss: 0.0128 - val_rpn_bbox_loss: 0.1185 - val_mrcnn_class_loss: 0.0413 - val_mrcnn_bbox_loss: 0.0874 - val_mrcnn_mask_loss: 0.1702\n",
      "Epoch 52/100\n",
      "\n",
      "Epoch 00052: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.5521 - rpn_class_loss: 0.0151 - rpn_bbox_loss: 0.1432 - mrcnn_class_loss: 0.0655 - mrcnn_bbox_loss: 0.1123 - mrcnn_mask_loss: 0.2160 - val_loss: 0.4230 - val_rpn_class_loss: 0.0168 - val_rpn_bbox_loss: 0.1178 - val_mrcnn_class_loss: 0.0269 - val_mrcnn_bbox_loss: 0.0869 - val_mrcnn_mask_loss: 0.1746\n",
      "Epoch 53/100\n",
      "\n",
      "Epoch 00053: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 782s 2s/step - loss: 0.5410 - rpn_class_loss: 0.0145 - rpn_bbox_loss: 0.1427 - mrcnn_class_loss: 0.0623 - mrcnn_bbox_loss: 0.1097 - mrcnn_mask_loss: 0.2119 - val_loss: 0.3975 - val_rpn_class_loss: 0.0126 - val_rpn_bbox_loss: 0.1086 - val_mrcnn_class_loss: 0.0303 - val_mrcnn_bbox_loss: 0.0712 - val_mrcnn_mask_loss: 0.1747\n",
      "Epoch 54/100\n",
      "\n",
      "Epoch 00054: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 781s 2s/step - loss: 0.5508 - rpn_class_loss: 0.0153 - rpn_bbox_loss: 0.1475 - mrcnn_class_loss: 0.0632 - mrcnn_bbox_loss: 0.1111 - mrcnn_mask_loss: 0.2137 - val_loss: 0.5774 - val_rpn_class_loss: 0.0218 - val_rpn_bbox_loss: 0.1749 - val_mrcnn_class_loss: 0.0407 - val_mrcnn_bbox_loss: 0.1258 - val_mrcnn_mask_loss: 0.2142\n",
      "Epoch 55/100\n",
      "\n",
      "Epoch 00055: LearningRateScheduler reducing learning rate to 0.001.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "343/343 [==============================] - 782s 2s/step - loss: 0.5477 - rpn_class_loss: 0.0159 - rpn_bbox_loss: 0.1450 - mrcnn_class_loss: 0.0649 - mrcnn_bbox_loss: 0.1093 - mrcnn_mask_loss: 0.2126 - val_loss: 0.4624 - val_rpn_class_loss: 0.0225 - val_rpn_bbox_loss: 0.1524 - val_mrcnn_class_loss: 0.0203 - val_mrcnn_bbox_loss: 0.0820 - val_mrcnn_mask_loss: 0.1852\n",
      "Epoch 56/100\n",
      "\n",
      "Epoch 00056: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 784s 2s/step - loss: 0.5433 - rpn_class_loss: 0.0152 - rpn_bbox_loss: 0.1406 - mrcnn_class_loss: 0.0627 - mrcnn_bbox_loss: 0.1106 - mrcnn_mask_loss: 0.2142 - val_loss: 0.4958 - val_rpn_class_loss: 0.0166 - val_rpn_bbox_loss: 0.1424 - val_mrcnn_class_loss: 0.0188 - val_mrcnn_bbox_loss: 0.1050 - val_mrcnn_mask_loss: 0.2130\n",
      "Epoch 57/100\n",
      "\n",
      "Epoch 00057: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 781s 2s/step - loss: 0.5497 - rpn_class_loss: 0.0150 - rpn_bbox_loss: 0.1459 - mrcnn_class_loss: 0.0634 - mrcnn_bbox_loss: 0.1114 - mrcnn_mask_loss: 0.2139 - val_loss: 0.4580 - val_rpn_class_loss: 0.0222 - val_rpn_bbox_loss: 0.1490 - val_mrcnn_class_loss: 0.0320 - val_mrcnn_bbox_loss: 0.0744 - val_mrcnn_mask_loss: 0.1803\n",
      "Epoch 58/100\n",
      "\n",
      "Epoch 00058: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.5419 - rpn_class_loss: 0.0153 - rpn_bbox_loss: 0.1450 - mrcnn_class_loss: 0.0586 - mrcnn_bbox_loss: 0.1097 - mrcnn_mask_loss: 0.2133 - val_loss: 0.4784 - val_rpn_class_loss: 0.0185 - val_rpn_bbox_loss: 0.1374 - val_mrcnn_class_loss: 0.0292 - val_mrcnn_bbox_loss: 0.1071 - val_mrcnn_mask_loss: 0.1863\n",
      "Epoch 59/100\n",
      "\n",
      "Epoch 00059: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5226 - rpn_class_loss: 0.0138 - rpn_bbox_loss: 0.1352 - mrcnn_class_loss: 0.0605 - mrcnn_bbox_loss: 0.1045 - mrcnn_mask_loss: 0.2085 - val_loss: 0.5349 - val_rpn_class_loss: 0.0157 - val_rpn_bbox_loss: 0.1434 - val_mrcnn_class_loss: 0.0526 - val_mrcnn_bbox_loss: 0.1253 - val_mrcnn_mask_loss: 0.1979\n",
      "Epoch 60/100\n",
      "\n",
      "Epoch 00060: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5423 - rpn_class_loss: 0.0154 - rpn_bbox_loss: 0.1394 - mrcnn_class_loss: 0.0619 - mrcnn_bbox_loss: 0.1105 - mrcnn_mask_loss: 0.2152 - val_loss: 0.4784 - val_rpn_class_loss: 0.0211 - val_rpn_bbox_loss: 0.1649 - val_mrcnn_class_loss: 0.0218 - val_mrcnn_bbox_loss: 0.0795 - val_mrcnn_mask_loss: 0.1910\n",
      "Epoch 61/100\n",
      "\n",
      "Epoch 00061: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 779s 2s/step - loss: 0.5166 - rpn_class_loss: 0.0137 - rpn_bbox_loss: 0.1324 - mrcnn_class_loss: 0.0584 - mrcnn_bbox_loss: 0.1037 - mrcnn_mask_loss: 0.2084 - val_loss: 0.4575 - val_rpn_class_loss: 0.0179 - val_rpn_bbox_loss: 0.1377 - val_mrcnn_class_loss: 0.0341 - val_mrcnn_bbox_loss: 0.0802 - val_mrcnn_mask_loss: 0.1876\n",
      "Epoch 62/100\n",
      "\n",
      "Epoch 00062: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 780s 2s/step - loss: 0.5199 - rpn_class_loss: 0.0150 - rpn_bbox_loss: 0.1342 - mrcnn_class_loss: 0.0577 - mrcnn_bbox_loss: 0.1028 - mrcnn_mask_loss: 0.2102 - val_loss: 0.5259 - val_rpn_class_loss: 0.0237 - val_rpn_bbox_loss: 0.1832 - val_mrcnn_class_loss: 0.0375 - val_mrcnn_bbox_loss: 0.0935 - val_mrcnn_mask_loss: 0.1880\n",
      "Epoch 63/100\n",
      "\n",
      "Epoch 00063: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5282 - rpn_class_loss: 0.0152 - rpn_bbox_loss: 0.1352 - mrcnn_class_loss: 0.0633 - mrcnn_bbox_loss: 0.1053 - mrcnn_mask_loss: 0.2093 - val_loss: 0.4254 - val_rpn_class_loss: 0.0116 - val_rpn_bbox_loss: 0.1120 - val_mrcnn_class_loss: 0.0180 - val_mrcnn_bbox_loss: 0.0921 - val_mrcnn_mask_loss: 0.1917\n",
      "Epoch 64/100\n",
      "\n",
      "Epoch 00064: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.5348 - rpn_class_loss: 0.0145 - rpn_bbox_loss: 0.1389 - mrcnn_class_loss: 0.0577 - mrcnn_bbox_loss: 0.1078 - mrcnn_mask_loss: 0.2159 - val_loss: 0.4451 - val_rpn_class_loss: 0.0150 - val_rpn_bbox_loss: 0.1253 - val_mrcnn_class_loss: 0.0285 - val_mrcnn_bbox_loss: 0.0960 - val_mrcnn_mask_loss: 0.1803\n",
      "Epoch 65/100\n",
      "\n",
      "Epoch 00065: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5205 - rpn_class_loss: 0.0133 - rpn_bbox_loss: 0.1351 - mrcnn_class_loss: 0.0568 - mrcnn_bbox_loss: 0.1052 - mrcnn_mask_loss: 0.2100 - val_loss: 0.4124 - val_rpn_class_loss: 0.0142 - val_rpn_bbox_loss: 0.1241 - val_mrcnn_class_loss: 0.0223 - val_mrcnn_bbox_loss: 0.0698 - val_mrcnn_mask_loss: 0.1821\n",
      "Epoch 66/100\n",
      "\n",
      "Epoch 00066: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.5274 - rpn_class_loss: 0.0146 - rpn_bbox_loss: 0.1341 - mrcnn_class_loss: 0.0602 - mrcnn_bbox_loss: 0.1058 - mrcnn_mask_loss: 0.2127 - val_loss: 0.4471 - val_rpn_class_loss: 0.0168 - val_rpn_bbox_loss: 0.1176 - val_mrcnn_class_loss: 0.0302 - val_mrcnn_bbox_loss: 0.0917 - val_mrcnn_mask_loss: 0.1908\n",
      "Epoch 67/100\n",
      "\n",
      "Epoch 00067: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 775s 2s/step - loss: 0.5382 - rpn_class_loss: 0.0150 - rpn_bbox_loss: 0.1370 - mrcnn_class_loss: 0.0614 - mrcnn_bbox_loss: 0.1085 - mrcnn_mask_loss: 0.2164 - val_loss: 0.4850 - val_rpn_class_loss: 0.0222 - val_rpn_bbox_loss: 0.1662 - val_mrcnn_class_loss: 0.0181 - val_mrcnn_bbox_loss: 0.0847 - val_mrcnn_mask_loss: 0.1939\n",
      "Epoch 68/100\n",
      "\n",
      "Epoch 00068: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.5053 - rpn_class_loss: 0.0133 - rpn_bbox_loss: 0.1315 - mrcnn_class_loss: 0.0552 - mrcnn_bbox_loss: 0.1002 - mrcnn_mask_loss: 0.2051 - val_loss: 0.5038 - val_rpn_class_loss: 0.0214 - val_rpn_bbox_loss: 0.1623 - val_mrcnn_class_loss: 0.0403 - val_mrcnn_bbox_loss: 0.0918 - val_mrcnn_mask_loss: 0.1880\n",
      "Epoch 69/100\n",
      "\n",
      "Epoch 00069: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 776s 2s/step - loss: 0.5187 - rpn_class_loss: 0.0143 - rpn_bbox_loss: 0.1338 - mrcnn_class_loss: 0.0571 - mrcnn_bbox_loss: 0.1036 - mrcnn_mask_loss: 0.2099 - val_loss: 0.5269 - val_rpn_class_loss: 0.0230 - val_rpn_bbox_loss: 0.1760 - val_mrcnn_class_loss: 0.0246 - val_mrcnn_bbox_loss: 0.1066 - val_mrcnn_mask_loss: 0.1967\n",
      "Epoch 70/100\n",
      "\n",
      "Epoch 00070: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 778s 2s/step - loss: 0.5103 - rpn_class_loss: 0.0128 - rpn_bbox_loss: 0.1292 - mrcnn_class_loss: 0.0559 - mrcnn_bbox_loss: 0.1015 - mrcnn_mask_loss: 0.2109 - val_loss: 0.4514 - val_rpn_class_loss: 0.0127 - val_rpn_bbox_loss: 0.1455 - val_mrcnn_class_loss: 0.0258 - val_mrcnn_bbox_loss: 0.0777 - val_mrcnn_mask_loss: 0.1896\n",
      "Epoch 71/100\n",
      "\n",
      "Epoch 00071: LearningRateScheduler reducing learning rate to 0.001.\n",
      "343/343 [==============================] - 777s 2s/step - loss: 0.5147 - rpn_class_loss: 0.0137 - rpn_bbox_loss: 0.1304 - mrcnn_class_loss: 0.0563 - mrcnn_bbox_loss: 0.1028 - mrcnn_mask_loss: 0.2115 - val_loss: 0.4798 - val_rpn_class_loss: 0.0136 - val_rpn_bbox_loss: 0.1297 - val_mrcnn_class_loss: 0.0354 - val_mrcnn_bbox_loss: 0.0992 - val_mrcnn_mask_loss: 0.2019\n",
      "Epoch 72/100\n",
      "\n",
      "Epoch 00072: LearningRateScheduler reducing learning rate to 0.001.\n",
      "304/343 [=========================>....] - ETA: 1:28 - loss: 0.5155 - rpn_class_loss: 0.0131 - rpn_bbox_loss: 0.1328 - mrcnn_class_loss: 0.0562 - mrcnn_bbox_loss: 0.1030 - mrcnn_mask_loss: 0.2104"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Process Process-3:\n",
      "Process Process-2:\n",
      "Traceback (most recent call last):\n",
      "  File \"/usr/lib/python3.5/multiprocessing/process.py\", line 249, in _bootstrap\n",
      "    self.run()\n",
      "  File \"/usr/lib/python3.5/multiprocessing/process.py\", line 93, in run\n",
      "    self._target(*self._args, **self._kwargs)\n",
      "  File \"/usr/local/lib/python3.5/dist-packages/keras/utils/data_utils.py\", line 680, in _data_generator_task\n",
      "    time.sleep(self.wait_time)\n",
      "KeyboardInterrupt\n",
      "Traceback (most recent call last):\n",
      "  File \"/usr/lib/python3.5/multiprocessing/process.py\", line 249, in _bootstrap\n",
      "    self.run()\n",
      "  File \"/usr/lib/python3.5/multiprocessing/process.py\", line 93, in run\n",
      "    self._target(*self._args, **self._kwargs)\n",
      "  File \"/usr/local/lib/python3.5/dist-packages/keras/utils/data_utils.py\", line 678, in _data_generator_task\n",
      "    self.queue.put((True, generator_output))\n",
      "  File \"<string>\", line 2, in put\n",
      "  File \"/usr/lib/python3.5/multiprocessing/managers.py\", line 717, in _callmethod\n",
      "    kind, result = conn.recv()\n",
      "  File \"/usr/lib/python3.5/multiprocessing/connection.py\", line 250, in recv\n",
      "    buf = self._recv_bytes()\n",
      "  File \"/usr/lib/python3.5/multiprocessing/connection.py\", line 407, in _recv_bytes\n",
      "    buf = self._recv(4)\n",
      "  File \"/usr/lib/python3.5/multiprocessing/connection.py\", line 379, in _recv\n",
      "    chunk = read(handle, remaining)\n",
      "KeyboardInterrupt\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-29-c954d2a8f872>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     38\u001b[0m                 \u001b[0mlearning_rate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mconfig\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mLEARNING_RATE\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     39\u001b[0m                 \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mepochs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mtrain_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 40\u001b[0;31m                 layers=\"all\", clbcks=[])\n\u001b[0m",
      "\u001b[0;32m/data/khyeh/Mask_RCNN/model.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, train_dataset, val_dataset, learning_rate, epochs, layers, clbcks)\u001b[0m\n\u001b[1;32m   2323\u001b[0m             \u001b[0mmax_queue_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2324\u001b[0m             \u001b[0mworkers\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mworkers\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2325\u001b[0;31m             \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2326\u001b[0m         )\n\u001b[1;32m   2327\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mepoch\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/legacy/interfaces.py\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     89\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     90\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 91\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     92\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     93\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2222\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2223\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2224\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2225\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2226\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1881\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1882\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1883\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1884\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1885\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2476\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2477\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2478\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2479\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2480\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    903\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    904\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 905\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    906\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    907\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1138\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1139\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1140\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1141\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1142\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1319\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1320\u001b[0m       return self._do_call(_run_fn, feeds, fetches, targets, options,\n\u001b[0;32m-> 1321\u001b[0;31m                            run_metadata)\n\u001b[0m\u001b[1;32m   1322\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1323\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1325\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1326\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1327\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1328\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1329\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1310\u001b[0m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_extend_graph\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1311\u001b[0m       return self._call_tf_sessionrun(\n\u001b[0;32m-> 1312\u001b[0;31m           options, feed_dict, fetch_list, target_list, run_metadata)\n\u001b[0m\u001b[1;32m   1313\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1314\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_call_tf_sessionrun\u001b[0;34m(self, options, feed_dict, fetch_list, target_list, run_metadata)\u001b[0m\n\u001b[1;32m   1418\u001b[0m         return tf_session.TF_Run(\n\u001b[1;32m   1419\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1420\u001b[0;31m             status, run_metadata)\n\u001b[0m\u001b[1;32m   1421\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1422\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_call_tf_sessionprun\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "if IS_DEVELOP:\n",
    "    epochs = [10, 10]\n",
    "    lr_shrink_period = [5, 5]\n",
    "    inital_lr = 0.01\n",
    "else:\n",
    "    epochs = [100, 200]\n",
    "    lr_shrink_period = [100, 150]\n",
    "    inital_lr = 0.001\n",
    "    \n",
    "for i in range(1):\n",
    "    train_index = i\n",
    "    \n",
    "    dataset_train, dataset_val = datasets[train_index]\n",
    "    \n",
    "    config.LEARNING_RATE = inital_lr\n",
    "    config.LR_SHRINK_PERIOD = lr_shrink_period[train_index]\n",
    "    config.STEPS_PER_EPOCH = len(dataset_train.image_ids) // config.IMAGES_PER_GPU\n",
    "    \n",
    "    model = modellib.MaskRCNN(mode=\"training\", config=config,\n",
    "                          model_dir=MODEL_DIR)\n",
    "    \n",
    "    init_with = \"coco\"  # imagenet, coco, or last\n",
    "\n",
    "    if init_with == \"imagenet\":\n",
    "        model.load_weights(model.get_imagenet_weights(), by_name=True)\n",
    "    elif init_with == \"coco\":\n",
    "        # Load weights trained on MS COCO, but skip layers that\n",
    "        # are different due to the different number of classes\n",
    "        # See README for instructions to download the COCO weights\n",
    "        model.load_weights(COCO_MODEL_PATH, by_name=True,\n",
    "                           exclude=[\"mrcnn_class_logits\", \"mrcnn_bbox_fc\", \n",
    "                                    \"mrcnn_bbox\", \"mrcnn_mask\"])\n",
    "    elif init_with == \"last\":\n",
    "        # Load the last model you trained and continue training\n",
    "        model.load_weights(model.find_last()[1], by_name=True)\n",
    "        \n",
    "    model.train(dataset_train, dataset_val, \n",
    "                learning_rate=config.LEARNING_RATE,\n",
    "                epochs=epochs[train_index], \n",
    "                layers=\"all\", clbcks=[])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's train and investigate the learning curves for all the mask-rcnn losses in the `charts` and `channels`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}