Kaivalya3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from skimage import io, morphology, img_as_bool, segmentation\n",
    "from scipy import ndimage as ndi\n",
    "from scipy.ndimage.morphology import binary_fill_holes\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "img_shape = (64,64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def pad_resize(img):\n",
    "    top = int((224 - img.shape[0])/2)\n",
    "    left = int((224 - img.shape[1])/2)\n",
    "    bottom = 224 - img.shape[0] - top\n",
    "    right = 224 - img.shape[1] - left\n",
    "    img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=255)\n",
    "    img = img/255.\n",
    "    img = cv2.resize(img, img_shape) #KADD\n",
    "    return img\n",
    "\n",
    "def skeletonize(img):\n",
    "    size = np.size(img)\n",
    "    skel = np.zeros(img.shape,np.uint8)\n",
    "    img = cv2.bitwise_not(img)\n",
    "#     element = cv2.getStructuringElement(cv2.MORPH_CROSS,(1,1))\n",
    "#     done = 0\n",
    "#     while( done < 1 ):\n",
    "#         eroded = cv2.erode(img,element)\n",
    "#         temp = cv2.dilate(eroded,element)\n",
    "#         temp = cv2.subtract(img,temp)\n",
    "#         skel = cv2.bitwise_or(skel,temp)\n",
    "#         img = eroded.copy()\n",
    "\n",
    "#         zeros = size - cv2.countNonZero(img)\n",
    "#         if zeros==size:#cv2.countNonZero(img) * 1 >= 0:#\n",
    "#             done += 1\n",
    "#     img = skel\n",
    "    kernel = np.ones((3,3),np.uint8)\n",
    "    erosion = cv2.erode(img,kernel,iterations = 2)\n",
    "    img = cv2.bitwise_not(erosion)\n",
    "    return img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 Done\n",
      "200 Done\n",
      "300 Done\n",
      "page3_6_0_2327_2417.png dropped!\n",
      "400 Done\n",
      "500 Done\n",
      "page5_7_8_2325_2367_2388.png dropped!\n",
      "page6_10_6_2325_2366_2380.png dropped!\n",
      "600 Done\n",
      "page2_4_20_2346_2366_2390.png dropped!\n",
      "700 Done\n",
      "page0_8_8_2330_2375_2379.png dropped!\n",
      "800 Done\n",
      "900 Done\n",
      "1000 Done\n",
      "page3_18_4_2332_2366_2379.png dropped!\n",
      "1100 Done\n",
      "page7_3_0_2319_2366_2367.png dropped!\n",
      "1200 Done\n",
      "1300 Done\n",
      "1400 Done\n",
      "1500 Done\n",
      "page6_1_1_2340_2375_2379.png dropped!\n",
      "1600 Done\n",
      "page0_9_15_2332_2375_2379.png dropped!\n",
      "page7_3_13_2319_2366_2367.png dropped!\n",
      "page3_12_0_2357_2364_2416.png dropped!\n",
      "1700 Done\n",
      "page7_16_10_2350_2417.png dropped!\n",
      "1800 Done\n"
     ]
    }
   ],
   "source": [
    "PATH = '../train_images_modified'\n",
    "images = []\n",
    "base_class = []\n",
    "matra_class = []\n",
    "dot_class = []\n",
    "total_class = []\n",
    "for filename in os.listdir(PATH):\n",
    "    if filename.endswith(\".png\"):\n",
    "        img = cv2.imread(os.path.join(PATH,filename),0)\n",
    "        blur = cv2.GaussianBlur(img,(9,9),0)# KADD\n",
    "        a,img = cv2.threshold(blur,127,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)#KEDIT\n",
    "#         kernel = np.ones((5,5),np.uint8)\n",
    "#         img = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel)\n",
    "#         plt.imshow(img, cmap='gray')\n",
    "#         plt.show()\n",
    "#        img = skeletonize(img)\n",
    "        img = pad_resize(img)\n",
    "        blur = cv2.GaussianBlur(img,(9,9),0)# KADD\n",
    "        char_arr = filename[:-4].split('_')[3:]\n",
    "        if(len(char_arr)>0):\n",
    "            char_arr = [int(i) for i in char_arr]\n",
    "            base = [i for i in char_arr if (i>=2308 and i<=2361) or (i==2384) or (i>=2392 and i<=2401) or (i>=2404 and i!=2416 and i!=2417)]  \n",
    "            matra = [i for i in char_arr if i>=2362 and i<=2391 and i!=2364 and i!=2362]\n",
    "            dot = [i for i in char_arr if i==2306 or i==2416 or i==2362 or i==2364]\n",
    "            if(len(matra)>1 or len(dot)>1 or len(base)>1 or (len(matra)==0 and len(dot)==0 and len(char_arr)==2)):\n",
    "                print(filename+\" dropped!\")\n",
    "            else:\n",
    "                images.append(img)\n",
    "                base_class.append(base[0])\n",
    "                if(len(matra)==1):\n",
    "                    matra_class.append(matra[0])\n",
    "                else:\n",
    "                    matra_class.append(0)\n",
    "\n",
    "                if len(dot)==1:\n",
    "                    if dot[0] == 2364:\n",
    "                        dot_class.append(2)\n",
    "                    else:\n",
    "                        dot_class.append(1)\n",
    "                else:\n",
    "                    dot_class.append(0)\n",
    "\n",
    "                if len(images)%100==0:\n",
    "                    print(\"{} Done\".format(len(images)))\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nICe+WYKc+GYJcuKbJej/ASLA9RZjDCvuAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc942bade48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1883\n",
      "1883\n"
     ]
    }
   ],
   "source": [
    "plt.imshow(images[10],\"gray\")\n",
    "plt.show()\n",
    "print(len(images))\n",
    "print(len(dot_class))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[2310, 2353, 2384, 2392, 2414, 2415]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(set(base_class))\n",
    "freq = {i:base_class.count(i) for i in base_class}\n",
    "[i for i in freq.keys() if freq[i]==1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1883, 64, 64)\n"
     ]
    }
   ],
   "source": [
    "images = np.array(images)\n",
    "print(images.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1883, 53)\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from keras.utils import np_utils\n",
    "\n",
    "le = LabelEncoder()\n",
    "le.fit(base_class)\n",
    "y_labeled = le.transform(base_class)\n",
    "y_train = np_utils.to_categorical(y_labeled)\n",
    "print(y_train.shape)\n",
    "#print(le)\n",
    "#le.get_params()\n",
    "#for i in range(len(topr)):\n",
    "#    if base_class[i] == 2405:\n",
    "#        print('{} {}'.format(topr[i], base_class[i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1883, 64, 64, 1)\n"
     ]
    }
   ],
   "source": [
    "x_train = np.reshape(images,(-1,img_shape[0],img_shape[1],1))\n",
    "print(x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense,Dropout,Conv2D,MaxPooling2D,Flatten\n",
    "from keras.optimizers import Adam\n",
    "\n",
    "base_model = Sequential()\n",
    "\n",
    "base_model.add(Conv2D(16,(3,3),input_shape = (img_shape[0],img_shape[1],1),activation = 'relu'))\n",
    "base_model.add(Conv2D(16,(3,3),activation = 'relu'))\n",
    "base_model.add(MaxPooling2D())\n",
    "base_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "base_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "base_model.add(MaxPooling2D())\n",
    "base_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "base_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "base_model.add(MaxPooling2D())\n",
    "base_model.add(MaxPooling2D())\n",
    "base_model.add(Dropout(0.5))\n",
    "base_model.add(Flatten())\n",
    "base_model.add(Dense(256,activation = 'sigmoid'))\n",
    "base_model.add(Dropout(0.25))\n",
    "base_model.add(Dense(len(set(base_class)),activation = 'softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 62, 62, 16)        160       \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 60, 60, 16)        2320      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 30, 30, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 28, 28, 32)        4640      \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 26, 26, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 13, 13, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 11, 11, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 9, 9, 64)          36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 2, 2, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 2, 2, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 256)               65792     \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 53)                13621     \n",
      "=================================================================\n",
      "Total params: 151,205\n",
      "Trainable params: 151,205\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import keras.backend as K\n",
    "\n",
    "def f1_score(theta):\n",
    "    def score(y_true, y_pred):\n",
    "\n",
    "        y_thresh = K.cast(K.greater(y_pred,theta),K.floatx())\n",
    "\n",
    "        true_pos =  K.sum(y_true * y_thresh)\n",
    "        false_pos = K.sum(y_true * (1. - y_thresh))\n",
    "        false_neg = K.sum((1. - y_true) * y_thresh)\n",
    "\n",
    "        precision = true_pos / (true_pos + false_pos)\n",
    "        recall = true_pos / (true_pos + false_neg)\n",
    "        \n",
    "        f1_score_val = 2 * (precision * recall) / (precision + recall)\n",
    "        return f1_score_val\n",
    "    return score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "base_model.compile(optimizer='adam',\n",
    "              loss='categorical_crossentropy', metrics=['categorical_accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# augmentation code here ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 3.5450 - categorical_accuracy: 0.0835 - val_loss: 3.3677 - val_categorical_accuracy: 0.0920\n",
      "Epoch 2/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 3.4018 - categorical_accuracy: 0.0842 - val_loss: 3.3592 - val_categorical_accuracy: 0.1027\n",
      "Epoch 3/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 3.3539 - categorical_accuracy: 0.0956 - val_loss: 3.2645 - val_categorical_accuracy: 0.1805\n",
      "Epoch 4/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 3.1274 - categorical_accuracy: 0.2033 - val_loss: 2.8279 - val_categorical_accuracy: 0.2619\n",
      "Epoch 5/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 2.6923 - categorical_accuracy: 0.3209 - val_loss: 2.2722 - val_categorical_accuracy: 0.4212\n",
      "Epoch 6/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 2.3683 - categorical_accuracy: 0.3778 - val_loss: 2.0453 - val_categorical_accuracy: 0.4619\n",
      "Epoch 7/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 2.1124 - categorical_accuracy: 0.4689 - val_loss: 1.8042 - val_categorical_accuracy: 0.5168\n",
      "Epoch 8/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 1.9028 - categorical_accuracy: 0.4848 - val_loss: 1.6041 - val_categorical_accuracy: 0.5717\n",
      "Epoch 9/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 1.6803 - categorical_accuracy: 0.5584 - val_loss: 1.4129 - val_categorical_accuracy: 0.6301\n",
      "Epoch 10/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 1.5187 - categorical_accuracy: 0.6017 - val_loss: 1.2462 - val_categorical_accuracy: 0.6602\n",
      "Epoch 11/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 1.3866 - categorical_accuracy: 0.6351 - val_loss: 1.2591 - val_categorical_accuracy: 0.6655\n",
      "Epoch 12/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 1.2781 - categorical_accuracy: 0.6487 - val_loss: 1.0752 - val_categorical_accuracy: 0.7097\n",
      "Epoch 13/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 1.1677 - categorical_accuracy: 0.6737 - val_loss: 1.0262 - val_categorical_accuracy: 0.7221\n",
      "Epoch 14/30\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 1.0982 - categorical_accuracy: 0.6950 - val_loss: 0.9533 - val_categorical_accuracy: 0.7469\n",
      "Epoch 15/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 1.0422 - categorical_accuracy: 0.7086 - val_loss: 0.8857 - val_categorical_accuracy: 0.7575\n",
      "Epoch 16/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.9574 - categorical_accuracy: 0.7337 - val_loss: 0.8676 - val_categorical_accuracy: 0.7664\n",
      "Epoch 17/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.9105 - categorical_accuracy: 0.7602 - val_loss: 0.8483 - val_categorical_accuracy: 0.7628\n",
      "Epoch 18/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.8678 - categorical_accuracy: 0.7534 - val_loss: 0.8235 - val_categorical_accuracy: 0.7805\n",
      "Epoch 19/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.8231 - categorical_accuracy: 0.7678 - val_loss: 0.7912 - val_categorical_accuracy: 0.7894\n",
      "Epoch 20/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.7494 - categorical_accuracy: 0.7914 - val_loss: 0.7642 - val_categorical_accuracy: 0.8088\n",
      "Epoch 21/30\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.7055 - categorical_accuracy: 0.8096 - val_loss: 0.7453 - val_categorical_accuracy: 0.7947\n",
      "Epoch 22/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.7193 - categorical_accuracy: 0.7974 - val_loss: 0.7296 - val_categorical_accuracy: 0.8018\n",
      "Epoch 23/30\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.6671 - categorical_accuracy: 0.8209 - val_loss: 0.7113 - val_categorical_accuracy: 0.8248\n",
      "Epoch 24/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.6491 - categorical_accuracy: 0.8156 - val_loss: 0.7230 - val_categorical_accuracy: 0.8035\n",
      "Epoch 25/30\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.6156 - categorical_accuracy: 0.8270 - val_loss: 0.6936 - val_categorical_accuracy: 0.8195\n",
      "Epoch 26/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.5880 - categorical_accuracy: 0.8285 - val_loss: 0.6814 - val_categorical_accuracy: 0.8442\n",
      "Epoch 27/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.5790 - categorical_accuracy: 0.8407 - val_loss: 0.7061 - val_categorical_accuracy: 0.8195\n",
      "Epoch 28/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.5092 - categorical_accuracy: 0.8619 - val_loss: 0.6387 - val_categorical_accuracy: 0.8336\n",
      "Epoch 29/30\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.4970 - categorical_accuracy: 0.8520 - val_loss: 0.6551 - val_categorical_accuracy: 0.8336\n",
      "Epoch 30/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.4750 - categorical_accuracy: 0.8566 - val_loss: 0.6657 - val_categorical_accuracy: 0.8442\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fc9348cb438>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_model.fit(x_train,y_train,epochs=30,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/10\n",
      "1318/1318 [==============================] - 15s 12ms/step - loss: 0.4207 - categorical_accuracy: 0.8832 - val_loss: 0.6386 - val_categorical_accuracy: 0.8460\n",
      "Epoch 2/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.4341 - categorical_accuracy: 0.8786 - val_loss: 0.6326 - val_categorical_accuracy: 0.8531\n",
      "Epoch 3/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.3942 - categorical_accuracy: 0.8991 - val_loss: 0.6281 - val_categorical_accuracy: 0.8442\n",
      "Epoch 4/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.4108 - categorical_accuracy: 0.8832 - val_loss: 0.6347 - val_categorical_accuracy: 0.8531\n",
      "Epoch 5/10\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.3861 - categorical_accuracy: 0.8907 - val_loss: 0.6406 - val_categorical_accuracy: 0.8496\n",
      "Epoch 6/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.3652 - categorical_accuracy: 0.8983 - val_loss: 0.6329 - val_categorical_accuracy: 0.8513\n",
      "Epoch 7/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.3830 - categorical_accuracy: 0.8930 - val_loss: 0.6339 - val_categorical_accuracy: 0.8460\n",
      "Epoch 8/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.3807 - categorical_accuracy: 0.8983 - val_loss: 0.6310 - val_categorical_accuracy: 0.8513\n",
      "Epoch 9/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.3571 - categorical_accuracy: 0.8953 - val_loss: 0.6249 - val_categorical_accuracy: 0.8584\n",
      "Epoch 10/10\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.3685 - categorical_accuracy: 0.8938 - val_loss: 0.6236 - val_categorical_accuracy: 0.8566\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fc93098ff98>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_adam = keras.optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "base_model.compile(optimizer=new_adam,\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "base_model.fit(x_train,y_train,epochs=10,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/30\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.2793 - categorical_accuracy: 0.9219 - val_loss: 0.5969 - val_categorical_accuracy: 0.8407\n",
      "Epoch 2/30\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.2636 - categorical_accuracy: 0.9264 - val_loss: 0.5898 - val_categorical_accuracy: 0.8442\n",
      "Epoch 3/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2675 - categorical_accuracy: 0.9302 - val_loss: 0.5908 - val_categorical_accuracy: 0.8389\n",
      "Epoch 4/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2585 - categorical_accuracy: 0.9401 - val_loss: 0.5933 - val_categorical_accuracy: 0.8442\n",
      "Epoch 5/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2633 - categorical_accuracy: 0.9325 - val_loss: 0.5965 - val_categorical_accuracy: 0.8460\n",
      "Epoch 6/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2695 - categorical_accuracy: 0.9234 - val_loss: 0.5934 - val_categorical_accuracy: 0.8425\n",
      "Epoch 7/30\n",
      "1318/1318 [==============================] - 15s 12ms/step - loss: 0.2536 - categorical_accuracy: 0.9355 - val_loss: 0.5912 - val_categorical_accuracy: 0.8425\n",
      "Epoch 8/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2611 - categorical_accuracy: 0.9370 - val_loss: 0.5876 - val_categorical_accuracy: 0.8425\n",
      "Epoch 9/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2549 - categorical_accuracy: 0.9347 - val_loss: 0.5818 - val_categorical_accuracy: 0.8425\n",
      "Epoch 10/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2336 - categorical_accuracy: 0.9408 - val_loss: 0.5812 - val_categorical_accuracy: 0.8442\n",
      "Epoch 11/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2713 - categorical_accuracy: 0.9287 - val_loss: 0.5784 - val_categorical_accuracy: 0.8460\n",
      "Epoch 12/30\n",
      "1318/1318 [==============================] - 15s 12ms/step - loss: 0.2658 - categorical_accuracy: 0.9165 - val_loss: 0.5840 - val_categorical_accuracy: 0.8442\n",
      "Epoch 13/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2498 - categorical_accuracy: 0.9332 - val_loss: 0.5882 - val_categorical_accuracy: 0.8425\n",
      "Epoch 14/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2590 - categorical_accuracy: 0.9302 - val_loss: 0.5857 - val_categorical_accuracy: 0.8442\n",
      "Epoch 15/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2215 - categorical_accuracy: 0.9446 - val_loss: 0.5890 - val_categorical_accuracy: 0.8425\n",
      "Epoch 16/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2523 - categorical_accuracy: 0.9340 - val_loss: 0.5870 - val_categorical_accuracy: 0.8425\n",
      "Epoch 17/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2411 - categorical_accuracy: 0.9378 - val_loss: 0.5777 - val_categorical_accuracy: 0.8478\n",
      "Epoch 18/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2346 - categorical_accuracy: 0.9423 - val_loss: 0.5762 - val_categorical_accuracy: 0.8460\n",
      "Epoch 19/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2481 - categorical_accuracy: 0.9287 - val_loss: 0.5760 - val_categorical_accuracy: 0.8478\n",
      "Epoch 20/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2319 - categorical_accuracy: 0.9454 - val_loss: 0.5738 - val_categorical_accuracy: 0.8460\n",
      "Epoch 21/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2361 - categorical_accuracy: 0.9363 - val_loss: 0.5765 - val_categorical_accuracy: 0.8460\n",
      "Epoch 22/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2203 - categorical_accuracy: 0.9454 - val_loss: 0.5773 - val_categorical_accuracy: 0.8442\n",
      "Epoch 23/30\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2333 - categorical_accuracy: 0.9416 - val_loss: 0.5752 - val_categorical_accuracy: 0.8460\n",
      "Epoch 24/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2214 - categorical_accuracy: 0.9408 - val_loss: 0.5743 - val_categorical_accuracy: 0.8442\n",
      "Epoch 25/30\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2173 - categorical_accuracy: 0.9446 - val_loss: 0.5813 - val_categorical_accuracy: 0.8513\n",
      "Epoch 26/30\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.2158 - categorical_accuracy: 0.9454 - val_loss: 0.5773 - val_categorical_accuracy: 0.8442\n",
      "Epoch 27/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2188 - categorical_accuracy: 0.9492 - val_loss: 0.5808 - val_categorical_accuracy: 0.8442\n",
      "Epoch 28/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2175 - categorical_accuracy: 0.9401 - val_loss: 0.5784 - val_categorical_accuracy: 0.8460\n",
      "Epoch 29/30\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.2249 - categorical_accuracy: 0.9347 - val_loss: 0.5757 - val_categorical_accuracy: 0.8442\n",
      "Epoch 30/30\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.2199 - categorical_accuracy: 0.9454 - val_loss: 0.5741 - val_categorical_accuracy: 0.8460\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f7cd07c7278>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_adam = keras.optimizers.Adam(lr=0.00005, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "base_model.compile(optimizer=new_adam,\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "base_model.fit(x_train,y_train,epochs=30,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "confusion matrix for (53, 53) features (base):\n",
      "[[ 1  0  0 ...,  0  0  0]\n",
      " [ 0 13  0 ...,  0  0  0]\n",
      " [ 0  1 29 ...,  0  0  0]\n",
      " ..., \n",
      " [ 0  0  0 ...,  0  0  0]\n",
      " [ 0  1  0 ...,  0  5  0]\n",
      " [ 0  0  0 ...,  0  0  0]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc92fbe9208>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABCIAAAReCAYAAAACMZHpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc92fbbae80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "y_pred = base_model.predict_classes(x_train)\n",
    "y_original = np.argmax(y_train, axis=1)\n",
    "\n",
    "cm = confusion_matrix(y_original, y_pred)\n",
    "print('confusion matrix for {} features (base):'.format(cm.shape))\n",
    "print(cm)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(20,20))\n",
    "sns.heatmap(cm, cmap=sns.light_palette('purple', n_colors=500, as_cmap=True), annot=True, linewidths = 5, ax=ax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DOT CLASS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_train_dot = np_utils.to_categorical(dot_class)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/10\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.3286 - categorical_accuracy: 0.9173 - val_loss: 0.3455 - val_categorical_accuracy: 0.9133\n",
      "Epoch 2/10\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.2883 - categorical_accuracy: 0.9340 - val_loss: 0.3437 - val_categorical_accuracy: 0.9133\n",
      "Epoch 3/10\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.2945 - categorical_accuracy: 0.9340 - val_loss: 0.3500 - val_categorical_accuracy: 0.9133\n",
      "Epoch 4/10\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.2811 - categorical_accuracy: 0.9340 - val_loss: 0.3471 - val_categorical_accuracy: 0.9133\n",
      "Epoch 5/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2894 - categorical_accuracy: 0.9340 - val_loss: 0.3661 - val_categorical_accuracy: 0.9133\n",
      "Epoch 6/10\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.2607 - categorical_accuracy: 0.9340 - val_loss: 0.3632 - val_categorical_accuracy: 0.9133\n",
      "Epoch 7/10\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.2631 - categorical_accuracy: 0.9340 - val_loss: 0.3222 - val_categorical_accuracy: 0.9133\n",
      "Epoch 8/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2383 - categorical_accuracy: 0.9340 - val_loss: 0.3114 - val_categorical_accuracy: 0.9133\n",
      "Epoch 9/10\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.2274 - categorical_accuracy: 0.9325 - val_loss: 0.3002 - val_categorical_accuracy: 0.9133\n",
      "Epoch 10/10\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.2184 - categorical_accuracy: 0.9385 - val_loss: 0.2716 - val_categorical_accuracy: 0.9133\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fc9075ea390>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dot_model = Sequential()\n",
    "\n",
    "dot_model.add(Conv2D(16,(3,3),input_shape = (img_shape[0],img_shape[1],1),activation = 'relu'))\n",
    "dot_model.add(Conv2D(16,(3,3),activation = 'relu'))\n",
    "dot_model.add(MaxPooling2D())\n",
    "dot_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "dot_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "dot_model.add(MaxPooling2D())\n",
    "dot_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "dot_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "dot_model.add(MaxPooling2D())\n",
    "dot_model.add(Flatten())\n",
    "dot_model.add(Dense(64,activation = 'sigmoid'))\n",
    "dot_model.add(Dropout(0.5))\n",
    "dot_model.add(Dense(len(set(dot_class)),activation = 'softmax'))\n",
    "\n",
    "dot_model.compile(optimizer='adam',\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "\n",
    "dot_model.fit(x_train,y_train_dot,epochs=10,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/5\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.1919 - categorical_accuracy: 0.9370 - val_loss: 0.2505 - val_categorical_accuracy: 0.9239\n",
      "Epoch 2/5\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.1771 - categorical_accuracy: 0.9423 - val_loss: 0.2391 - val_categorical_accuracy: 0.9327\n",
      "Epoch 3/5\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.1693 - categorical_accuracy: 0.9431 - val_loss: 0.2263 - val_categorical_accuracy: 0.9345\n",
      "Epoch 4/5\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.1494 - categorical_accuracy: 0.9446 - val_loss: 0.2225 - val_categorical_accuracy: 0.9345\n",
      "Epoch 5/5\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.1418 - categorical_accuracy: 0.9446 - val_loss: 0.2289 - val_categorical_accuracy: 0.9327\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fc9077f1e10>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_adam = keras.optimizers.Adam(lr=0.0005, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "dot_model.compile(optimizer=new_adam,\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "dot_model.fit(x_train,y_train_dot,epochs=5,batch_size=64,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### confusion matrix:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "confusion matrix for (3, 3) features (dot):\n",
      "[[1737   10    0]\n",
      " [  68   49    0]\n",
      " [  19    0    0]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc90692b208>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAccAAAFKCAYAAABo0pS0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHJ9ZvXKBAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc906927978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "y_pred = dot_model.predict_classes(x_train)\n",
    "y_original = np.argmax(y_train_dot, axis=1)\n",
    "\n",
    "cm = confusion_matrix(y_original, y_pred)\n",
    "print('confusion matrix for {} features (dot):'.format(cm.shape))\n",
    "print(cm)\n",
    "\n",
    "#fig, ax = plt.subplots(figsize=(20,20))\n",
    "sns.heatmap(cm, cmap=sns.light_palette('purple', n_colors=500, as_cmap=True), annot=True, linewidths = 5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MATRA CLASS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1883, 16)\n"
     ]
    }
   ],
   "source": [
    "le_matra = LabelEncoder()\n",
    "le_matra.fit(matra_class)\n",
    "y_labeled_matra = le_matra.transform(matra_class)\n",
    "y_train_matra = np_utils.to_categorical(y_labeled_matra)\n",
    "\n",
    "print(y_train_matra.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1779 - categorical_accuracy: 0.4469 - val_loss: 0.1635 - val_categorical_accuracy: 0.4690\n",
      "Epoch 2/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.1537 - categorical_accuracy: 0.4894 - val_loss: 0.1284 - val_categorical_accuracy: 0.5522\n",
      "Epoch 3/35\n",
      "1318/1318 [==============================] - 18s 13ms/step - loss: 0.1251 - categorical_accuracy: 0.5903 - val_loss: 0.1182 - val_categorical_accuracy: 0.6230\n",
      "Epoch 4/35\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.1109 - categorical_accuracy: 0.6533 - val_loss: 0.0901 - val_categorical_accuracy: 0.7292\n",
      "Epoch 5/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0928 - categorical_accuracy: 0.7033 - val_loss: 0.0763 - val_categorical_accuracy: 0.7681\n",
      "Epoch 6/35\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.0818 - categorical_accuracy: 0.7489 - val_loss: 0.0697 - val_categorical_accuracy: 0.7912\n",
      "Epoch 7/35\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.0779 - categorical_accuracy: 0.7511 - val_loss: 0.0658 - val_categorical_accuracy: 0.8000\n",
      "Epoch 8/35\n",
      "1318/1318 [==============================] - 16s 13ms/step - loss: 0.0689 - categorical_accuracy: 0.7807 - val_loss: 0.0777 - val_categorical_accuracy: 0.7929\n",
      "Epoch 9/35\n",
      "1318/1318 [==============================] - 15s 12ms/step - loss: 0.0690 - categorical_accuracy: 0.7754 - val_loss: 0.0587 - val_categorical_accuracy: 0.8212\n",
      "Epoch 10/35\n",
      "1318/1318 [==============================] - 15s 12ms/step - loss: 0.0634 - categorical_accuracy: 0.7876 - val_loss: 0.0600 - val_categorical_accuracy: 0.8265\n",
      "Epoch 11/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.0618 - categorical_accuracy: 0.8073 - val_loss: 0.0596 - val_categorical_accuracy: 0.8195\n",
      "Epoch 12/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0600 - categorical_accuracy: 0.8149 - val_loss: 0.0573 - val_categorical_accuracy: 0.8354\n",
      "Epoch 13/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.0546 - categorical_accuracy: 0.8323 - val_loss: 0.0581 - val_categorical_accuracy: 0.8372\n",
      "Epoch 14/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0519 - categorical_accuracy: 0.8445 - val_loss: 0.0505 - val_categorical_accuracy: 0.8602\n",
      "Epoch 15/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0500 - categorical_accuracy: 0.8498 - val_loss: 0.0489 - val_categorical_accuracy: 0.8690\n",
      "Epoch 16/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0482 - categorical_accuracy: 0.8634 - val_loss: 0.0472 - val_categorical_accuracy: 0.8779\n",
      "Epoch 17/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0488 - categorical_accuracy: 0.8551 - val_loss: 0.0460 - val_categorical_accuracy: 0.8726\n",
      "Epoch 18/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0446 - categorical_accuracy: 0.8710 - val_loss: 0.0462 - val_categorical_accuracy: 0.8779\n",
      "Epoch 19/35\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.0418 - categorical_accuracy: 0.8778 - val_loss: 0.0453 - val_categorical_accuracy: 0.8779\n",
      "Epoch 20/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0408 - categorical_accuracy: 0.8703 - val_loss: 0.0442 - val_categorical_accuracy: 0.8850\n",
      "Epoch 21/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0366 - categorical_accuracy: 0.8991 - val_loss: 0.0414 - val_categorical_accuracy: 0.8938\n",
      "Epoch 22/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.0353 - categorical_accuracy: 0.9021 - val_loss: 0.0422 - val_categorical_accuracy: 0.8920\n",
      "Epoch 23/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0339 - categorical_accuracy: 0.9067 - val_loss: 0.0419 - val_categorical_accuracy: 0.8956\n",
      "Epoch 24/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0323 - categorical_accuracy: 0.9044 - val_loss: 0.0441 - val_categorical_accuracy: 0.8973\n",
      "Epoch 25/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0297 - categorical_accuracy: 0.9196 - val_loss: 0.0413 - val_categorical_accuracy: 0.9009\n",
      "Epoch 26/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0314 - categorical_accuracy: 0.9074 - val_loss: 0.0384 - val_categorical_accuracy: 0.9115\n",
      "Epoch 27/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0328 - categorical_accuracy: 0.9067 - val_loss: 0.0419 - val_categorical_accuracy: 0.9009\n",
      "Epoch 28/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0281 - categorical_accuracy: 0.9211 - val_loss: 0.0428 - val_categorical_accuracy: 0.9027\n",
      "Epoch 29/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0288 - categorical_accuracy: 0.9203 - val_loss: 0.0468 - val_categorical_accuracy: 0.8885\n",
      "Epoch 30/35\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.0292 - categorical_accuracy: 0.9181 - val_loss: 0.0408 - val_categorical_accuracy: 0.9080\n",
      "Epoch 31/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0274 - categorical_accuracy: 0.9241 - val_loss: 0.0393 - val_categorical_accuracy: 0.8991\n",
      "Epoch 32/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0248 - categorical_accuracy: 0.9347 - val_loss: 0.0395 - val_categorical_accuracy: 0.9133\n",
      "Epoch 33/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.0241 - categorical_accuracy: 0.9325 - val_loss: 0.0433 - val_categorical_accuracy: 0.9115\n",
      "Epoch 34/35\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.0225 - categorical_accuracy: 0.9378 - val_loss: 0.0364 - val_categorical_accuracy: 0.9204\n",
      "Epoch 35/35\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.0208 - categorical_accuracy: 0.9454 - val_loss: 0.0387 - val_categorical_accuracy: 0.9168\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f7cb9222828>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "matra_model = Sequential()\n",
    "\n",
    "matra_model.add(Conv2D(16,(3,3),input_shape = (img_shape[0],img_shape[1],1),activation = 'relu'))\n",
    "matra_model.add(Conv2D(16,(3,3),activation = 'relu'))\n",
    "matra_model.add(MaxPooling2D())\n",
    "matra_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "matra_model.add(Conv2D(32,(3,3),activation = 'relu'))\n",
    "matra_model.add(MaxPooling2D())\n",
    "matra_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "matra_model.add(Conv2D(64,(3,3),activation = 'relu'))\n",
    "matra_model.add(MaxPooling2D())\n",
    "matra_model.add(MaxPooling2D())\n",
    "matra_model.add(Dropout(0.5))\n",
    "matra_model.add(Flatten())\n",
    "matra_model.add(Dense(128,activation = 'sigmoid'))\n",
    "matra_model.add(Dropout(0.25))\n",
    "matra_model.add(Dense(len(set(matra_class)),activation = 'softmax'))\n",
    "\n",
    "matra_model.compile(optimizer='adam',\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "\n",
    "matra_model.fit(x_train,y_train_matra,epochs=35,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1318 samples, validate on 565 samples\n",
      "Epoch 1/15\n",
      "1318/1318 [==============================] - 12s 9ms/step - loss: 0.1608 - categorical_accuracy: 0.9545 - val_loss: 0.3796 - val_categorical_accuracy: 0.9168\n",
      "Epoch 2/15\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.1682 - categorical_accuracy: 0.9431 - val_loss: 0.3748 - val_categorical_accuracy: 0.9186\n",
      "Epoch 3/15\n",
      "1318/1318 [==============================] - 13s 10ms/step - loss: 0.1570 - categorical_accuracy: 0.9484 - val_loss: 0.3781 - val_categorical_accuracy: 0.9186\n",
      "Epoch 4/15\n",
      "1318/1318 [==============================] - 14s 10ms/step - loss: 0.1580 - categorical_accuracy: 0.9530 - val_loss: 0.3848 - val_categorical_accuracy: 0.9168\n",
      "Epoch 5/15\n",
      "1318/1318 [==============================] - 15s 11ms/step - loss: 0.1593 - categorical_accuracy: 0.9575 - val_loss: 0.3815 - val_categorical_accuracy: 0.9168\n",
      "Epoch 6/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1722 - categorical_accuracy: 0.9575 - val_loss: 0.3812 - val_categorical_accuracy: 0.9168\n",
      "Epoch 7/15\n",
      "1318/1318 [==============================] - 17s 13ms/step - loss: 0.1580 - categorical_accuracy: 0.9590 - val_loss: 0.3806 - val_categorical_accuracy: 0.9150\n",
      "Epoch 8/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1618 - categorical_accuracy: 0.9568 - val_loss: 0.3753 - val_categorical_accuracy: 0.9186\n",
      "Epoch 9/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1425 - categorical_accuracy: 0.9552 - val_loss: 0.3764 - val_categorical_accuracy: 0.9186\n",
      "Epoch 10/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1565 - categorical_accuracy: 0.9552 - val_loss: 0.3834 - val_categorical_accuracy: 0.9168\n",
      "Epoch 11/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1488 - categorical_accuracy: 0.9522 - val_loss: 0.3773 - val_categorical_accuracy: 0.9168\n",
      "Epoch 12/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1534 - categorical_accuracy: 0.9560 - val_loss: 0.3768 - val_categorical_accuracy: 0.9204\n",
      "Epoch 13/15\n",
      "1318/1318 [==============================] - 16s 12ms/step - loss: 0.1476 - categorical_accuracy: 0.9590 - val_loss: 0.3765 - val_categorical_accuracy: 0.9204\n",
      "Epoch 14/15\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.1576 - categorical_accuracy: 0.9560 - val_loss: 0.3805 - val_categorical_accuracy: 0.9204\n",
      "Epoch 15/15\n",
      "1318/1318 [==============================] - 14s 11ms/step - loss: 0.1532 - categorical_accuracy: 0.9560 - val_loss: 0.3785 - val_categorical_accuracy: 0.9204\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f7ca6e13438>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_adam = keras.optimizers.Adam(lr=0.00005, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "matra_model.compile(optimizer=new_adam,\n",
    "              loss='categorical_crossentropy',\n",
    "              metrics=['categorical_accuracy'])\n",
    "matra_model.fit(x_train,y_train_matra,epochs=20,batch_size=32,validation_split=0.3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1883/1883 [==============================] - 6s 3ms/step\n",
      "confusion matrix for (16, 16) features (matra):\n",
      "[[892   0   4   2   0   3   2   0   0   0   0   1   0   0   0   0]\n",
      " [  0   0   7   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  5   0 284   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0 179   0   0   0   0   0   1   0   0   0   0   0   0]\n",
      " [  1   0   2   1 137   0   0   0   0   0   1   0   0   0   0   0]\n",
      " [  3   0   0   0   0  39   0   0   0   0   0   0   0   0   0   0]\n",
      " [ 11   0   0   0   1   0  16   0   0   0   0   0   0   0   0   0]\n",
      " [  1   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  2   0   0   1   0   0   0   0  81   0   0   0   0   0   0   0]\n",
      " [  0   0   0   4   0   0   0   0   1  78   0   3   0   0   0   0]\n",
      " [  0   0   0   0   6   0   0   0   3   0  46   1   0   0   0   0]\n",
      " [  0   0   1   0   0   0   0   0   0   4   1  53   0   0   0   0]\n",
      " [  0   0   1   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  1   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  1   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
      " [  2   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f7cb85a5f60>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABCQAAARiCAYAAAB8lyZ4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7cb85a5940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "\n",
    "y_pred = matra_model.predict_classes(x_train)\n",
    "y_original = np.argmax(y_train_matra, axis=1)\n",
    "\n",
    "cm = confusion_matrix(y_original, y_pred)\n",
    "print('confusion matrix for {} features (matra):'.format(cm.shape))\n",
    "print(cm)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(20,20))\n",
    "sns.heatmap(cm, cmap=sns.light_palette('purple', n_colors=500, as_cmap=True), annot=True, linewidths = 5, ax=ax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dumping Models to Disk"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dot_model.save('models/dot_model3k.model')\n",
    "matra_model.save('models/matra_model3k.model')\n",
    "base_model.save('models/base_model3k.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.models import load_model\n",
    "\n",
    "dot_model2 = load_model('models/dot_model3k.model')\n",
    "matra_model2 = load_model('models/matra_model3k.model')\n",
    "base_model2 = load_model('models/base_model3k.model')\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Half Characters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2310, 2311, 2312, 2313, 2315, 2319, 2325, 2327, 2328, 2330, 2331,\n",
       "       2332, 2334, 2335, 2336, 2337, 2338, 2340, 2341, 2342, 2343, 2344,\n",
       "       2346, 2347, 2348, 2349, 2350, 2351, 2352, 2353, 2354, 2357, 2358,\n",
       "       2359, 2360, 2361, 2384, 2392, 2399, 2404, 2405, 2406, 2407, 2408,\n",
       "       2409, 2410, 2411, 2412, 2413, 2414, 2415, 2424, 2429])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([   0, 2363, 2366, 2367, 2368, 2369, 2370, 2372, 2375, 2376, 2379,\n",
       "       2380, 2382, 2384, 2387, 2390])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "le_matra.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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
}