Lab-siamese-nn-Recognize icebergs and ships-Questions.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.6.8"},"colab":{"name":"Lab-siamese-nn-Recognize icebergs and ships-Questions.ipynb","version":"0.3.2","provenance":[],"collapsed_sections":[]},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"_cell_guid":"608e43e3-8699-4299-9d18-9e9a6a3dd9bf","_uuid":"434ffd54c774edd4e8b851ab0a4b6675425e46da","id":"7V2_29yj57XG","colab_type":"text"},"source":["# Siamese Neural Networks\n","Recognize images of icebergs from images of ships. \n","These images were taken from space with a Sentinel-4 satellite. The goal was to create the most accurate neural network to differentiate them. However, neural networks usually only work well with A LOT of data, and I constrained you guys to 1000 examples to train on. That seems like a lot, but a CONVENTIONAL neural network needs more data. However, we are going to make a special kind of neural network: *Siamese Network*. This special network is used for face recognition and few-shot learning(learning from few examples). Let's get started by importing the usual liraries:"]},{"cell_type":"code","metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","id":"E6hbtL5u57XH","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"5109645c-41c5-4471-fd90-4f85862f95c6","executionInfo":{"status":"ok","timestamp":1566577246639,"user_tz":-330,"elapsed":4259,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["from keras.layers import Input, Conv2D, Lambda, merge, Dense, Flatten,MaxPooling2D,Activation, Dropout\n","from keras.models import Model, Sequential\n","from keras.regularizers import l2\n","from keras import backend as K\n","from keras.optimizers import Adam\n","from skimage.io import imshow\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import pandas as pd\n","import random"],"execution_count":1,"outputs":[{"output_type":"stream","text":["Using TensorFlow backend.\n"],"name":"stderr"}]},{"cell_type":"markdown","metadata":{"_cell_guid":"adbffb0a-1784-4595-b3f2-2ea930abc31f","_uuid":"b9200b733a896e4b6f261122fe9134c00955cc91","id":"OVGk4ygY57XL","colab_type":"text"},"source":["## Load the Data\n"]},{"cell_type":"code","metadata":{"id":"gavlVNfLR_D-","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":121},"outputId":"879d864f-2092-494b-9cfc-522ffbdd3f17","executionInfo":{"status":"ok","timestamp":1566577303836,"user_tz":-330,"elapsed":37567,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["from google.colab import drive\n","drive.mount('/content/drive/')"],"execution_count":2,"outputs":[{"output_type":"stream","text":["Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n","\n","Enter your authorization code:\n","··········\n","Mounted at /content/drive/\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"CnMkp3BbSDUI","colab_type":"code","colab":{}},"source":["project_path = \"/content/drive/My Drive/AdvanceComputerVission/\""],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"qFaydzmnSdPO","colab_type":"code","colab":{}},"source":["import os\n","os.chdir(project_path)\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"wZR4JsgvSfQI","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"9f1959d5-fb33-4dcf-a756-8a345e37f32b","executionInfo":{"status":"ok","timestamp":1566577394700,"user_tz":-330,"elapsed":2820,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["pwd"],"execution_count":6,"outputs":[{"output_type":"execute_result","data":{"text/plain":["'/content/drive/My Drive/AdvanceComputerVission'"]},"metadata":{"tags":[]},"execution_count":6}]},{"cell_type":"code","metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","id":"_fOCBf8k57XM","colab_type":"code","outputId":"3316ec32-ae25-45ff-8432-9374e740016b","colab":{"base_uri":"https://localhost:8080/","height":34},"executionInfo":{"status":"ok","timestamp":1566577413849,"user_tz":-330,"elapsed":9155,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["npz = np.load(project_path+'input_data.npz')\n","X_train = npz['X_train']\n","Y_train = npz['Y_train']\n","del npz\n","print ('We have {} examples to work with'.format(Y_train.shape[0]-1000))"],"execution_count":8,"outputs":[{"output_type":"stream","text":["We have 3113 examples to work with\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"_cell_guid":"ed28bc84-1a8a-472f-98ae-9553fe6d1097","_uuid":"b2e2306b8b16ae7f4f9b63249d1f2ff6e29559d3","id":"lcRmyOTU57XS","colab_type":"text"},"source":["## Check the data\n","\n","Load some of the images to observe how your data look like."]},{"cell_type":"code","metadata":{"id":"ollkonjHTY-T","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"d3d43af6-d1ce-4f93-bf70-807e12828f5e","executionInfo":{"status":"ok","timestamp":1566577636501,"user_tz":-330,"elapsed":871,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["X_train.shape"],"execution_count":10,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(4113, 75, 75, 3)"]},"metadata":{"tags":[]},"execution_count":10}]},{"cell_type":"code","metadata":{"id":"fpaqaeTCTuTE","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"0f8be05d-2742-4282-85ac-cbfaf1dc7508","executionInfo":{"status":"ok","timestamp":1566577728843,"user_tz":-330,"elapsed":863,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["Y_train.shape"],"execution_count":12,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(4113,)"]},"metadata":{"tags":[]},"execution_count":12}]},{"cell_type":"code","metadata":{"id":"zzcO_B7I57XS","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":314},"outputId":"fa05421b-f30e-4ee0-8ff2-195882387bf0","executionInfo":{"status":"ok","timestamp":1566577825652,"user_tz":-330,"elapsed":1130,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["index=100\n","imshow(np.squeeze(X_train[index,:,:,2]))\n","plt.show()\n","labels = ['Ship','Iceberg']\n","print ('This is:',labels[int(Y_train[index])])"],"execution_count":18,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAUcAAAEYCAYAAADPkTRJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBAJD8P8Dga8NnaZaB4wAAAAASUVORK5CYII=\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["This is: Ship\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"ygCwiArZUNOJ","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":314},"outputId":"fd1eaea4-7317-44e7-8d09-9587af9f1b1b","executionInfo":{"status":"ok","timestamp":1566577847740,"user_tz":-330,"elapsed":1160,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["index=101\n","imshow(np.squeeze(X_train[index,:,:,2]))\n","plt.show()\n","labels = ['Ship','Iceberg']\n","print ('This is:',labels[int(Y_train[index])])"],"execution_count":19,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAUcAAAEYCAYAAADPkTRJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n","text/plain":["<Figure size 432x288 with 2 Axes>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["This is: Iceberg\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"TpC6GlG957XU","colab_type":"text"},"source":["# Creating the structure of Siamese Network"]},{"cell_type":"markdown","metadata":{"id":"AfPxWyPX57XV","colab_type":"raw"},"source":["You just need to break it into some steps:\n","1. 2 Inputs for each images\n","2. Creating a network which both images will go through individually\n","3. Couple the network to each input\n","4. Calculate the L1 distance between them. Just (x1,y1)-(x2,y2)\n","5. 1 Added layer that will say 1 if they are the same and 0 if they are different"]},{"cell_type":"markdown","metadata":{"id":"TLNLPe-E57XW","colab_type":"text"},"source":["# Create two Inputs for each image - left_input and right_input\n","Hint - use Input Function in Keras and keep the shape to be 75,75,3"]},{"cell_type":"code","metadata":{"id":"c-szk7ld57XX","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":121},"outputId":"2626f389-d04f-4ae6-c0c3-814217df6479","executionInfo":{"status":"ok","timestamp":1566577927952,"user_tz":-330,"elapsed":987,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["left_input = Input((75,75,3))\n","right_input = Input((75,75,3))"],"execution_count":21,"outputs":[{"output_type":"stream","text":["WARNING: Logging before flag parsing goes to stderr.\n","W0823 16:32:07.369656 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n","\n","W0823 16:32:07.409168 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n","\n"],"name":"stderr"}]},{"cell_type":"markdown","metadata":{"id":"N9Y8sX_f57Xa","colab_type":"text"},"source":["# Create a base Network in which both images will go through individually.\n","\n","Hint - Use Sequential Model in Keras with convolution layers, Maxpooling Layers, Flatten Layer and Denser Layer with appropriate activation functions.\n","Name the network as base_network"]},{"cell_type":"code","metadata":{"id":"ybSDpqyq57Xb","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":104},"outputId":"37b301dc-d6db-4ec5-d5a5-0de5285f1b95","executionInfo":{"status":"ok","timestamp":1566578477491,"user_tz":-330,"elapsed":841,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["base_network = Sequential([\n","    Conv2D(5,3,input_shape=(75,75,3)),\n","    Activation('relu'),\n","    MaxPooling2D(),\n","    Conv2D(5,3),\n","    Activation('relu'),\n","    MaxPooling2D(),\n","    Conv2D(7,2),\n","    Activation('relu'),\n","    MaxPooling2D(),\n","    Conv2D(7,2),\n","    Activation('relu'),\n","    MaxPooling2D(),\n","    Flatten(),\n","    Dense(18),\n","    Activation('sigmoid')\n","])"],"execution_count":23,"outputs":[{"output_type":"stream","text":["W0823 16:41:16.965641 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n","\n","W0823 16:41:16.995998 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n","\n"],"name":"stderr"}]},{"cell_type":"code","metadata":{"id":"vs8yUyLe57Xd","colab_type":"code","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"PFrJuNz657Xh","colab_type":"text"},"source":["# Process the left and right inputs to the same Base Network(base_network)"]},{"cell_type":"code","metadata":{"id":"zWhCMi4H57Xi","colab_type":"code","colab":{}},"source":["processed_a=base_network(left_input)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"R5dk3fjS57Xn","colab_type":"code","colab":{}},"source":["processed_b=base_network(right_input)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"3q3oxKSs57Xq","colab_type":"text"},"source":["# Get the L1 Distance layer between the 2 processed encodings (processed_a and processed_b)\n","Hint - You can use Lambda function for this."]},{"cell_type":"code","metadata":{"id":"gYAWnnq257Xr","colab_type":"code","colab":{}},"source":["L1_layer=Lambda(lambda tensor:K.abs(tensor[0] - tensor[1]))"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"73bGeoV757X2","colab_type":"code","colab":{}},"source":["L1_distance = L1_layer([processed_a, processed_b])"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"LbHdgP_B57X4","colab_type":"text"},"source":["# Add the prediction layer in the end and create the Siamese Network using Model function."]},{"cell_type":"code","metadata":{"id":"KUh2H0dc57X5","colab_type":"code","colab":{}},"source":["prediction = Dense(1,activation='sigmoid')(L1_distance)\n","siamese_net = Model(inputs=[left_input,right_input],outputs=prediction)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Q4hmG5Gf57X8","colab_type":"text"},"source":["# Define the Optimizer Adam and Compile the Siamese Model."]},{"cell_type":"code","metadata":{"id":"T7mWyBkU57X9","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":154},"outputId":"da3c5805-62fe-4af1-ff0c-6186a818de88","executionInfo":{"status":"ok","timestamp":1566578983201,"user_tz":-330,"elapsed":871,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["optimizer = Adam(0.001, decay=2.5e-4)\n","\n","siamese_net.compile(loss=\"binary_crossentropy\",optimizer=optimizer,metrics=['accuracy'])"],"execution_count":33,"outputs":[{"output_type":"stream","text":["W0823 16:49:42.571673 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n","\n","W0823 16:49:42.577130 139656131454848 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3376: The name tf.log is deprecated. Please use tf.math.log instead.\n","\n","W0823 16:49:42.581012 139656131454848 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/nn_impl.py:180: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n","Instructions for updating:\n","Use tf.where in 2.0, which has the same broadcast rule as np.where\n"],"name":"stderr"}]},{"cell_type":"markdown","metadata":{"_cell_guid":"dbba8e74-6b9c-4c84-96ed-55137265280b","_uuid":"28e3843cae1a192bb14eb04cc1fe1705ef69f405","id":"xwPdtLKL57X_","colab_type":"text"},"source":["# Show the Summary of your Siamese Network"]},{"cell_type":"code","metadata":{"id":"XL19TZa557YA","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":336},"outputId":"6b2d738c-345b-44ed-d746-a5cac3a685f3","executionInfo":{"status":"ok","timestamp":1566579015384,"user_tz":-330,"elapsed":843,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["siamese_net.summary()"],"execution_count":34,"outputs":[{"output_type":"stream","text":["__________________________________________________________________________________________________\n","Layer (type)                    Output Shape         Param #     Connected to                     \n","==================================================================================================\n","input_1 (InputLayer)            (None, 75, 75, 3)    0                                            \n","__________________________________________________________________________________________________\n","input_2 (InputLayer)            (None, 75, 75, 3)    0                                            \n","__________________________________________________________________________________________________\n","sequential_1 (Sequential)       (None, 18)           1872        input_1[0][0]                    \n","                                                                 input_2[0][0]                    \n","__________________________________________________________________________________________________\n","lambda_1 (Lambda)               (None, 18)           0           sequential_1[1][0]               \n","                                                                 sequential_1[3][0]               \n","__________________________________________________________________________________________________\n","dense_3 (Dense)                 (None, 1)            19          lambda_1[0][0]                   \n","==================================================================================================\n","Total params: 1,891\n","Trainable params: 1,891\n","Non-trainable params: 0\n","__________________________________________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"bfFLHiD457YC","colab_type":"text"},"source":["# Creating the Pair of Images to train on"]},{"cell_type":"markdown","metadata":{"_cell_guid":"7080234f-01f4-4df1-88a7-459746335c50","_uuid":"4553890c8130dc76076252a09290497a4afc27e6","id":"jDGMlDVW57YC","colab_type":"text"},"source":["It's not time to train though. We still have to create pairs of images to train on. There will be Positive(the same class) or Negative(different classes) for outputs. Let's construct this dataset. It is recommended that there are equal amounts positive and negative."]},{"cell_type":"code","metadata":{"_cell_guid":"32cb30db-144d-4f93-af05-6cb9a697c084","_uuid":"09a3e9d488acd34e99912c5f01474029e77637f5","id":"zWto2syW57YD","colab_type":"code","colab":{}},"source":["# First let's separate the dataset from 1 matrix to a list of matricies\n","image_list = np.split(X_train[:1000],1000)\n","label_list = np.split(Y_train[:1000],1000)\n","\n","left_input = []\n","right_input = []\n","targets = []\n","\n","#Number of pairs per image\n","pairs = 5\n","#Let's create the new dataset to train on\n","for i in range(len(label_list)):\n","    for _ in range(pairs):\n","        compare_to = i\n","        while compare_to == i: #Make sure it's not comparing to itself\n","            compare_to = random.randint(0,999)\n","        left_input.append(image_list[i])\n","        right_input.append(image_list[compare_to])\n","        if label_list[i] == label_list[compare_to]:# They are the same\n","            targets.append(1.)\n","        else:# Not the same\n","            targets.append(0.)\n","            \n","left_input = np.squeeze(np.array(left_input))\n","right_input = np.squeeze(np.array(right_input))\n","targets = np.squeeze(np.array(targets))\n","\n","iceimage = X_train[101]\n","test_left = []\n","test_right = []\n","test_targets = []\n","\n","for i in range(Y_train.shape[0]-1000):\n","    test_left.append(iceimage)\n","    test_right.append(X_train[i+1000])\n","    test_targets.append(Y_train[i+1000])\n","\n","test_left = np.squeeze(np.array(test_left))\n","test_right = np.squeeze(np.array(test_right))\n","test_targets = np.squeeze(np.array(test_targets))"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Tdabr1nu57YG","colab_type":"text"},"source":["Now we have a lot more examples.\n","Now we have pairs x 1000 examples to train the network on. \n","Each side will have an input of an image and the output will be one if they are the same and zero if not."]},{"cell_type":"markdown","metadata":{"id":"Xnc9wqjt57YH","colab_type":"text"},"source":["# Train the Siamese train on left and right input image pairs created and validate it by test pairs."]},{"cell_type":"code","metadata":{"id":"I8I4IL8k57YI","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":1000},"outputId":"2bf39a35-aae6-47a4-d0c1-813878454089","executionInfo":{"status":"ok","timestamp":1566579473689,"user_tz":-330,"elapsed":351930,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["siamese_net.fit([left_input,right_input], targets,\n","          batch_size=16,\n","          epochs=100,\n","          verbose=1,\n","          validation_data=([test_left,test_right],test_targets))"],"execution_count":38,"outputs":[{"output_type":"stream","text":["Train on 5000 samples, validate on 3113 samples\n","Epoch 1/100\n","5000/5000 [==============================] - 4s 739us/step - loss: 0.6251 - acc: 0.6624 - val_loss: 0.5911 - val_acc: 0.7343\n","Epoch 2/100\n","5000/5000 [==============================] - 4s 746us/step - loss: 0.6147 - acc: 0.6736 - val_loss: 0.6335 - val_acc: 0.6627\n","Epoch 3/100\n","5000/5000 [==============================] - 4s 732us/step - loss: 0.6064 - acc: 0.6820 - val_loss: 0.6340 - val_acc: 0.6756\n","Epoch 4/100\n","5000/5000 [==============================] - 4s 725us/step - loss: 0.5962 - acc: 0.6956 - val_loss: 0.7166 - val_acc: 0.5316\n","Epoch 5/100\n","5000/5000 [==============================] - 4s 702us/step - loss: 0.5849 - acc: 0.7024 - val_loss: 0.7842 - val_acc: 0.4709\n","Epoch 6/100\n","5000/5000 [==============================] - 4s 713us/step - loss: 0.5731 - acc: 0.7160 - val_loss: 0.5685 - val_acc: 0.7388\n","Epoch 7/100\n","5000/5000 [==============================] - 4s 718us/step - loss: 0.5610 - acc: 0.7234 - val_loss: 0.6466 - val_acc: 0.6515\n","Epoch 8/100\n","5000/5000 [==============================] - 4s 724us/step - loss: 0.5508 - acc: 0.7312 - val_loss: 0.5627 - val_acc: 0.7340\n","Epoch 9/100\n","5000/5000 [==============================] - 4s 724us/step - loss: 0.5380 - acc: 0.7416 - val_loss: 0.5544 - val_acc: 0.7437\n","Epoch 10/100\n","5000/5000 [==============================] - 4s 704us/step - loss: 0.5275 - acc: 0.7508 - val_loss: 0.6545 - val_acc: 0.6136\n","Epoch 11/100\n","5000/5000 [==============================] - 4s 705us/step - loss: 0.5166 - acc: 0.7622 - val_loss: 0.4697 - val_acc: 0.8198\n","Epoch 12/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.5057 - acc: 0.7694 - val_loss: 0.5432 - val_acc: 0.7633\n","Epoch 13/100\n","5000/5000 [==============================] - 4s 704us/step - loss: 0.4977 - acc: 0.7742 - val_loss: 0.4988 - val_acc: 0.8069\n","Epoch 14/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.4874 - acc: 0.7848 - val_loss: 0.5237 - val_acc: 0.7986\n","Epoch 15/100\n","5000/5000 [==============================] - 4s 726us/step - loss: 0.4787 - acc: 0.7870 - val_loss: 0.5226 - val_acc: 0.7957\n","Epoch 16/100\n","5000/5000 [==============================] - 4s 738us/step - loss: 0.4716 - acc: 0.7926 - val_loss: 0.4640 - val_acc: 0.8204\n","Epoch 17/100\n","5000/5000 [==============================] - 4s 707us/step - loss: 0.4618 - acc: 0.7994 - val_loss: 0.4873 - val_acc: 0.8156\n","Epoch 18/100\n","5000/5000 [==============================] - 4s 706us/step - loss: 0.4550 - acc: 0.8014 - val_loss: 0.4703 - val_acc: 0.8201\n","Epoch 19/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.4501 - acc: 0.8054 - val_loss: 0.4672 - val_acc: 0.8169\n","Epoch 20/100\n","5000/5000 [==============================] - 4s 704us/step - loss: 0.4418 - acc: 0.8120 - val_loss: 0.4346 - val_acc: 0.8201\n","Epoch 21/100\n","5000/5000 [==============================] - 4s 700us/step - loss: 0.4361 - acc: 0.8150 - val_loss: 0.4889 - val_acc: 0.8037\n","Epoch 22/100\n","5000/5000 [==============================] - 3s 695us/step - loss: 0.4329 - acc: 0.8184 - val_loss: 0.4615 - val_acc: 0.8137\n","Epoch 23/100\n","5000/5000 [==============================] - 3s 688us/step - loss: 0.4265 - acc: 0.8226 - val_loss: 0.4855 - val_acc: 0.8082\n","Epoch 24/100\n","5000/5000 [==============================] - 4s 707us/step - loss: 0.4232 - acc: 0.8232 - val_loss: 0.4496 - val_acc: 0.8114\n","Epoch 25/100\n","5000/5000 [==============================] - 3s 692us/step - loss: 0.4171 - acc: 0.8280 - val_loss: 0.5036 - val_acc: 0.7983\n","Epoch 26/100\n","5000/5000 [==============================] - 4s 702us/step - loss: 0.4148 - acc: 0.8298 - val_loss: 0.4634 - val_acc: 0.8069\n","Epoch 27/100\n","5000/5000 [==============================] - 3s 700us/step - loss: 0.4084 - acc: 0.8354 - val_loss: 0.4811 - val_acc: 0.8073\n","Epoch 28/100\n","5000/5000 [==============================] - 3s 696us/step - loss: 0.4064 - acc: 0.8378 - val_loss: 0.4637 - val_acc: 0.8063\n","Epoch 29/100\n","5000/5000 [==============================] - 3s 700us/step - loss: 0.4021 - acc: 0.8358 - val_loss: 0.4847 - val_acc: 0.8053\n","Epoch 30/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.3966 - acc: 0.8414 - val_loss: 0.4760 - val_acc: 0.8012\n","Epoch 31/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.3935 - acc: 0.8438 - val_loss: 0.4601 - val_acc: 0.8092\n","Epoch 32/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.3905 - acc: 0.8432 - val_loss: 0.4466 - val_acc: 0.8102\n","Epoch 33/100\n","5000/5000 [==============================] - 4s 700us/step - loss: 0.3886 - acc: 0.8458 - val_loss: 0.4564 - val_acc: 0.8079\n","Epoch 34/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.3824 - acc: 0.8510 - val_loss: 0.4763 - val_acc: 0.8021\n","Epoch 35/100\n","5000/5000 [==============================] - 4s 701us/step - loss: 0.3806 - acc: 0.8504 - val_loss: 0.4719 - val_acc: 0.8031\n","Epoch 36/100\n","5000/5000 [==============================] - 4s 710us/step - loss: 0.3776 - acc: 0.8514 - val_loss: 0.4734 - val_acc: 0.8012\n","Epoch 37/100\n","5000/5000 [==============================] - 4s 701us/step - loss: 0.3730 - acc: 0.8546 - val_loss: 0.4654 - val_acc: 0.8047\n","Epoch 38/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.3698 - acc: 0.8588 - val_loss: 0.4792 - val_acc: 0.7999\n","Epoch 39/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.3675 - acc: 0.8602 - val_loss: 0.4823 - val_acc: 0.7983\n","Epoch 40/100\n","5000/5000 [==============================] - 4s 708us/step - loss: 0.3635 - acc: 0.8604 - val_loss: 0.4673 - val_acc: 0.8066\n","Epoch 41/100\n","5000/5000 [==============================] - 3s 698us/step - loss: 0.3602 - acc: 0.8638 - val_loss: 0.4770 - val_acc: 0.8034\n","Epoch 42/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.3569 - acc: 0.8650 - val_loss: 0.4681 - val_acc: 0.8069\n","Epoch 43/100\n","5000/5000 [==============================] - 4s 720us/step - loss: 0.3556 - acc: 0.8662 - val_loss: 0.4966 - val_acc: 0.7967\n","Epoch 44/100\n","5000/5000 [==============================] - 3s 690us/step - loss: 0.3516 - acc: 0.8666 - val_loss: 0.4960 - val_acc: 0.7983\n","Epoch 45/100\n","5000/5000 [==============================] - 3s 699us/step - loss: 0.3501 - acc: 0.8690 - val_loss: 0.4962 - val_acc: 0.7996\n","Epoch 46/100\n","5000/5000 [==============================] - 3s 690us/step - loss: 0.3473 - acc: 0.8700 - val_loss: 0.5069 - val_acc: 0.7957\n","Epoch 47/100\n","5000/5000 [==============================] - 4s 707us/step - loss: 0.3441 - acc: 0.8718 - val_loss: 0.4956 - val_acc: 0.7999\n","Epoch 48/100\n","5000/5000 [==============================] - 3s 700us/step - loss: 0.3427 - acc: 0.8692 - val_loss: 0.5162 - val_acc: 0.7938\n","Epoch 49/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.3411 - acc: 0.8718 - val_loss: 0.5067 - val_acc: 0.7973\n","Epoch 50/100\n","5000/5000 [==============================] - 4s 700us/step - loss: 0.3388 - acc: 0.8724 - val_loss: 0.5343 - val_acc: 0.7832\n","Epoch 51/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.3361 - acc: 0.8734 - val_loss: 0.5107 - val_acc: 0.7976\n","Epoch 52/100\n","5000/5000 [==============================] - 4s 700us/step - loss: 0.3340 - acc: 0.8742 - val_loss: 0.4988 - val_acc: 0.8018\n","Epoch 53/100\n","5000/5000 [==============================] - 3s 699us/step - loss: 0.3316 - acc: 0.8764 - val_loss: 0.5189 - val_acc: 0.7944\n","Epoch 54/100\n","5000/5000 [==============================] - 4s 709us/step - loss: 0.3293 - acc: 0.8754 - val_loss: 0.5059 - val_acc: 0.7989\n","Epoch 55/100\n","5000/5000 [==============================] - 3s 694us/step - loss: 0.3279 - acc: 0.8762 - val_loss: 0.5350 - val_acc: 0.7893\n","Epoch 56/100\n","5000/5000 [==============================] - 3s 684us/step - loss: 0.3263 - acc: 0.8768 - val_loss: 0.5183 - val_acc: 0.7976\n","Epoch 57/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.3247 - acc: 0.8776 - val_loss: 0.5315 - val_acc: 0.7915\n","Epoch 58/100\n","5000/5000 [==============================] - 3s 696us/step - loss: 0.3220 - acc: 0.8804 - val_loss: 0.5188 - val_acc: 0.7979\n","Epoch 59/100\n","5000/5000 [==============================] - 3s 698us/step - loss: 0.3209 - acc: 0.8806 - val_loss: 0.5174 - val_acc: 0.7999\n","Epoch 60/100\n","5000/5000 [==============================] - 3s 699us/step - loss: 0.3190 - acc: 0.8810 - val_loss: 0.5259 - val_acc: 0.7970\n","Epoch 61/100\n","5000/5000 [==============================] - 4s 704us/step - loss: 0.3178 - acc: 0.8840 - val_loss: 0.5240 - val_acc: 0.7983\n","Epoch 62/100\n","5000/5000 [==============================] - 3s 691us/step - loss: 0.3155 - acc: 0.8818 - val_loss: 0.5182 - val_acc: 0.7979\n","Epoch 63/100\n","5000/5000 [==============================] - 4s 700us/step - loss: 0.3146 - acc: 0.8840 - val_loss: 0.5536 - val_acc: 0.7861\n","Epoch 64/100\n","5000/5000 [==============================] - 3s 684us/step - loss: 0.3130 - acc: 0.8848 - val_loss: 0.5455 - val_acc: 0.7909\n","Epoch 65/100\n","5000/5000 [==============================] - 3s 687us/step - loss: 0.3110 - acc: 0.8866 - val_loss: 0.5422 - val_acc: 0.7918\n","Epoch 66/100\n","5000/5000 [==============================] - 3s 693us/step - loss: 0.3093 - acc: 0.8872 - val_loss: 0.5382 - val_acc: 0.7973\n","Epoch 67/100\n","5000/5000 [==============================] - 4s 701us/step - loss: 0.3074 - acc: 0.8884 - val_loss: 0.5340 - val_acc: 0.7970\n","Epoch 68/100\n","5000/5000 [==============================] - 3s 696us/step - loss: 0.3069 - acc: 0.8858 - val_loss: 0.5303 - val_acc: 0.7973\n","Epoch 69/100\n","5000/5000 [==============================] - 3s 689us/step - loss: 0.3043 - acc: 0.8900 - val_loss: 0.5528 - val_acc: 0.7928\n","Epoch 70/100\n","5000/5000 [==============================] - 3s 688us/step - loss: 0.3036 - acc: 0.8894 - val_loss: 0.5291 - val_acc: 0.7963\n","Epoch 71/100\n","5000/5000 [==============================] - 3s 696us/step - loss: 0.3014 - acc: 0.8920 - val_loss: 0.5293 - val_acc: 0.7983\n","Epoch 72/100\n","5000/5000 [==============================] - 4s 704us/step - loss: 0.3005 - acc: 0.8914 - val_loss: 0.5189 - val_acc: 0.7983\n","Epoch 73/100\n","5000/5000 [==============================] - 3s 693us/step - loss: 0.2999 - acc: 0.8910 - val_loss: 0.5467 - val_acc: 0.7944\n","Epoch 74/100\n","5000/5000 [==============================] - 3s 687us/step - loss: 0.2979 - acc: 0.8918 - val_loss: 0.5332 - val_acc: 0.7983\n","Epoch 75/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.2961 - acc: 0.8948 - val_loss: 0.5506 - val_acc: 0.7947\n","Epoch 76/100\n","5000/5000 [==============================] - 3s 699us/step - loss: 0.2951 - acc: 0.8938 - val_loss: 0.5332 - val_acc: 0.7967\n","Epoch 77/100\n","5000/5000 [==============================] - 4s 703us/step - loss: 0.2936 - acc: 0.8938 - val_loss: 0.5533 - val_acc: 0.7957\n","Epoch 78/100\n","5000/5000 [==============================] - 4s 702us/step - loss: 0.2935 - acc: 0.8952 - val_loss: 0.5433 - val_acc: 0.7963\n","Epoch 79/100\n","5000/5000 [==============================] - 3s 694us/step - loss: 0.2908 - acc: 0.8972 - val_loss: 0.5473 - val_acc: 0.7944\n","Epoch 80/100\n","5000/5000 [==============================] - 3s 690us/step - loss: 0.2897 - acc: 0.8994 - val_loss: 0.5599 - val_acc: 0.7938\n","Epoch 81/100\n","5000/5000 [==============================] - 3s 682us/step - loss: 0.2884 - acc: 0.8988 - val_loss: 0.5596 - val_acc: 0.7954\n","Epoch 82/100\n","5000/5000 [==============================] - 3s 691us/step - loss: 0.2876 - acc: 0.8988 - val_loss: 0.5669 - val_acc: 0.7899\n","Epoch 83/100\n","5000/5000 [==============================] - 3s 694us/step - loss: 0.2857 - acc: 0.9010 - val_loss: 0.5766 - val_acc: 0.7889\n","Epoch 84/100\n","5000/5000 [==============================] - 3s 692us/step - loss: 0.2854 - acc: 0.8998 - val_loss: 0.5602 - val_acc: 0.7951\n","Epoch 85/100\n","5000/5000 [==============================] - 3s 683us/step - loss: 0.2845 - acc: 0.9002 - val_loss: 0.5521 - val_acc: 0.7963\n","Epoch 86/100\n","5000/5000 [==============================] - 3s 689us/step - loss: 0.2834 - acc: 0.9016 - val_loss: 0.5688 - val_acc: 0.7906\n","Epoch 87/100\n","5000/5000 [==============================] - 3s 693us/step - loss: 0.2829 - acc: 0.9020 - val_loss: 0.5496 - val_acc: 0.7963\n","Epoch 88/100\n","5000/5000 [==============================] - 3s 689us/step - loss: 0.2804 - acc: 0.9042 - val_loss: 0.5513 - val_acc: 0.7963\n","Epoch 89/100\n","5000/5000 [==============================] - 3s 694us/step - loss: 0.2804 - acc: 0.9036 - val_loss: 0.5645 - val_acc: 0.7909\n","Epoch 90/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.2782 - acc: 0.9044 - val_loss: 0.5679 - val_acc: 0.7938\n","Epoch 91/100\n","5000/5000 [==============================] - 4s 712us/step - loss: 0.2777 - acc: 0.9030 - val_loss: 0.5615 - val_acc: 0.7912\n","Epoch 92/100\n","5000/5000 [==============================] - 4s 717us/step - loss: 0.2769 - acc: 0.9054 - val_loss: 0.5774 - val_acc: 0.7889\n","Epoch 93/100\n","5000/5000 [==============================] - 3s 687us/step - loss: 0.2755 - acc: 0.9044 - val_loss: 0.5708 - val_acc: 0.7906\n","Epoch 94/100\n","5000/5000 [==============================] - 3s 694us/step - loss: 0.2750 - acc: 0.9060 - val_loss: 0.5747 - val_acc: 0.7877\n","Epoch 95/100\n","5000/5000 [==============================] - 3s 691us/step - loss: 0.2744 - acc: 0.9056 - val_loss: 0.5768 - val_acc: 0.7893\n","Epoch 96/100\n","5000/5000 [==============================] - 3s 696us/step - loss: 0.2723 - acc: 0.9070 - val_loss: 0.5420 - val_acc: 0.7986\n","Epoch 97/100\n","5000/5000 [==============================] - 3s 685us/step - loss: 0.2718 - acc: 0.9072 - val_loss: 0.5625 - val_acc: 0.7928\n","Epoch 98/100\n","5000/5000 [==============================] - 3s 688us/step - loss: 0.2718 - acc: 0.9068 - val_loss: 0.5612 - val_acc: 0.7954\n","Epoch 99/100\n","5000/5000 [==============================] - 3s 699us/step - loss: 0.2700 - acc: 0.9064 - val_loss: 0.5757 - val_acc: 0.7909\n","Epoch 100/100\n","5000/5000 [==============================] - 3s 697us/step - loss: 0.2689 - acc: 0.9082 - val_loss: 0.5685 - val_acc: 0.7944\n"],"name":"stdout"},{"output_type":"execute_result","data":{"text/plain":["<keras.callbacks.History at 0x7f03e0c2ce80>"]},"metadata":{"tags":[]},"execution_count":38}]},{"cell_type":"code","metadata":{"id":"5dufj0cFiPKX","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"d153d866-20b9-4841-d8b3-814ce9351f1c","executionInfo":{"status":"ok","timestamp":1566581952563,"user_tz":-330,"elapsed":834,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["X_train[1,1,1,1]"],"execution_count":54,"outputs":[{"output_type":"execute_result","data":{"text/plain":["0.011488241686355972"]},"metadata":{"tags":[]},"execution_count":54}]},{"cell_type":"code","metadata":{"id":"LZcPIOPL57YK","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"ebadf93e-e4de-4309-df80-f86c65deb62a","executionInfo":{"status":"ok","timestamp":1566582157127,"user_tz":-330,"elapsed":861,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh3.googleusercontent.com/a-/AAuE7mA_nHqFdfM_zFBpb-BYHYJwL7O3uKBW_MNkY6jW=s64","userId":"16774972310128518178"}}},"source":["siamese_net.predict([X_train[0].reshape(-1,75, 75, 3) ,X_train[1].reshape(-1,75,75, 3) ])"],"execution_count":62,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array([[0.88468]], dtype=float32)"]},"metadata":{"tags":[]},"execution_count":62}]},{"cell_type":"markdown","metadata":{"id":"fG7flfcKk836","colab_type":"text"},"source":["This is close to 1 so it could be a ICE BERG\n"]},{"cell_type":"code","metadata":{"id":"xucIAyv457YT","colab_type":"code","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"IZD9vxK957YY","colab_type":"code","colab":{}},"source":[""],"execution_count":0,"outputs":[]}]}