Digit_Sequence_Recognition.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Digit Sequence Recognizaton.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "572ZvIDaBXbO",
        "colab_type": "code",
        "outputId": "d3695c45-07e4-4add-c930-1df716daaabe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import random\n",
        "\n",
        "import numpy as np\n",
        "from PIL import Image\n",
        "import tensorflow as tf\n",
        "\n",
        "from keras.datasets import mnist\n",
        "from keras.utils import np_utils\n",
        "from keras.models import Sequential, Model\n",
        "from keras.layers import Dense, Dropout, Activation, Flatten, Input\n",
        "from keras.layers import Convolution2D, MaxPool2D\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8A_5aDW4CCqW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Loading MNIST Dataset from keras\n",
        "\n",
        "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2iXF65ZmCeOS",
        "colab_type": "code",
        "outputId": "341eafd2-e7dd-4048-bee4-9d1e1b7e704f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "print(f'Shape of training dataset: {np.shape(X_train)}')\n",
        "print(f'Shape of testing dataset: {np.shape(X_test)}')"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Shape of training dataset: (60000, 28, 28)\n",
            "Shape of testing dataset: (10000, 28, 28)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8V-uWTwnC6_o",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Creating Custom Dataset for digits in sequence\n",
        "\n",
        "def build_dataset(data, labels, dataset_size):\n",
        "  new_img_width = 64\n",
        "  new_img_height = 64\n",
        "\n",
        "  new_data = np.ndarray(shape=(dataset_size, new_img_height, new_img_width), dtype=np.float32)\n",
        "  new_labels = []\n",
        "\n",
        "  for i in range(0, dataset_size):\n",
        "    num_digits = random.randint(1, 5)\n",
        "\n",
        "    indices = [random.randint(0, len(data)-1) for _ in range(0, num_digits)]\n",
        "    new_image = np.hstack([X_train[index] for index in indices])\n",
        "    new_label = [y_train[index] for index in indices]\n",
        "\n",
        "    for j in range(0, 5-num_digits):\n",
        "      new_image = np.hstack([new_image, np.zeros((28, 28))])\n",
        "      new_label.append(10)\n",
        "\n",
        "    new_image = np.array(Image.fromarray(new_image).resize((64, 64)))\n",
        "    new_data[i,:,:] = new_image\n",
        "    new_labels.append(tuple(new_label))\n",
        "  \n",
        "  return new_data, new_labels\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TOV1fDy4G96O",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Building new Training and Testing Dataset\n",
        "\n",
        "X_new_train, y_new_train = build_dataset(X_train, y_train, 60000)\n",
        "X_new_test, y_new_test = build_dataset(X_train, y_train, 60000)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jaxIa9pGHR71",
        "colab_type": "code",
        "outputId": "c7d802a4-6b59-4275-a6f9-115f8c5fe819",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "source": [
        "plt.figure()\n",
        "plt.imshow(X_new_train[108], cmap='gray')\n",
        "print(y_new_train[108])\n",
        "\n",
        "# here 10 shows empty space"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(3, 0, 4, 7, 10)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WKPeAtTvLNZU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Conver labels to One-hot representation of shape (set_Size, digits, classes)\n",
        "#Possible Classes are 11 because there are 0-9 digits and another class to represent that no number was there hence 11 classes\n",
        "possible_classes = 11\n",
        "\n",
        "def convert_labels(labels):\n",
        "  dig0_arr = np.ndarray(shape=(len(labels),possible_classes))\n",
        "  dig1_arr = np.ndarray(shape=(len(labels),possible_classes))\n",
        "  dig2_arr = np.ndarray(shape=(len(labels),possible_classes))\n",
        "  dig3_arr = np.ndarray(shape=(len(labels),possible_classes)) \n",
        "  dig4_arr = np.ndarray(shape=(len(labels),possible_classes))\n",
        "  for index, label in enumerate(labels):\n",
        "    dig0_arr[index,:] = np_utils.to_categorical(label[0],possible_classes)\n",
        "    dig1_arr[index,:] = np_utils.to_categorical(label[1],possible_classes)\n",
        "    dig2_arr[index,:] = np_utils.to_categorical(label[2],possible_classes)\n",
        "    dig3_arr[index,:] = np_utils.to_categorical(label[3],possible_classes)\n",
        "    dig4_arr[index,:] = np_utils.to_categorical(label[4],possible_classes)\n",
        "\n",
        "  return [dig0_arr, dig1_arr, dig2_arr, dig3_arr, dig4_arr]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6UPFTHXUMOAu",
        "colab_type": "code",
        "outputId": "a9d3867f-d33a-482f-cdc6-33200ef61153",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "train_labels = convert_labels(y_new_train)\n",
        "test_labels = convert_labels(y_new_test)\n",
        "\n",
        "print(np.shape(train_labels))"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(5, 60000, 11)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C3KrYcloMbt2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Preprocessing Images\n",
        "\n",
        "def process_data(img_data):\n",
        "  img_data = img_data.reshape(len(img_data),64,64,1)\n",
        "  img_data = img_data.astype('float32')\n",
        "  img_data = img_data/255\n",
        "  return img_data"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "t0Xo98hBMpQC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_images = process_data(X_new_train)\n",
        "test_images = process_data(X_new_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EazS0iPJNSDn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Model Building\n",
        "\n",
        "batch_size = 128\n",
        "\n",
        "inputs = Input(shape=(64, 64, 1))\n",
        "\n",
        "cov = Convolution2D(32, 3, strides=3, padding='same', activation='relu')(inputs)\n",
        "cov = Convolution2D(32, 3, strides=3, padding='same', activation='relu')(cov)\n",
        "cov = MaxPool2D(pool_size=(2, 2))(cov)\n",
        "cov = Dropout(0.25)(cov)\n",
        "cov = Flatten()(cov)\n",
        "\n",
        "cov = Dense(128, activation='relu')(cov)\n",
        "cov = Dropout(0.5)(cov)\n",
        "\n",
        "c0 = Dense(11, activation='softmax')(cov)\n",
        "c1 = Dense(11, activation='softmax')(cov)\n",
        "c2 = Dense(11, activation='softmax')(cov)\n",
        "c3 = Dense(11, activation='softmax')(cov)\n",
        "c4 = Dense(11, activation='softmax')(cov)\n",
        "\n",
        "model = Model(inputs=inputs, outputs=[c0, c1, c2, c3, c4])\n",
        "\n",
        "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xuCudzksNVSX",
        "colab_type": "code",
        "outputId": "f30920c9-0507-4cfb-c7dd-6d52f40ade9e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 574
        }
      },
      "source": [
        "model.fit(train_images, train_labels, batch_size=batch_size, epochs=15, \n",
        "          verbose=1, validation_data=(test_images, test_labels))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 60000 samples, validate on 60000 samples\n",
            "Epoch 1/15\n",
            "60000/60000 [==============================] - 9s 144us/step - loss: 6.0966 - dense_2_loss: 1.6091 - dense_3_loss: 1.5507 - dense_4_loss: 1.3524 - dense_5_loss: 1.0145 - dense_6_loss: 0.5692 - dense_2_accuracy: 0.4319 - dense_3_accuracy: 0.4584 - dense_4_accuracy: 0.5345 - dense_5_accuracy: 0.6672 - dense_6_accuracy: 0.8243 - val_loss: 3.4131 - val_dense_2_loss: 0.7380 - val_dense_3_loss: 0.8486 - val_dense_4_loss: 0.8260 - val_dense_5_loss: 0.6520 - val_dense_6_loss: 0.3485 - val_dense_2_accuracy: 0.8056 - val_dense_3_accuracy: 0.7769 - val_dense_4_accuracy: 0.7806 - val_dense_5_accuracy: 0.8077 - val_dense_6_accuracy: 0.9025\n",
            "Epoch 2/15\n",
            "60000/60000 [==============================] - 7s 120us/step - loss: 4.0727 - dense_2_loss: 0.9809 - dense_3_loss: 1.0195 - dense_4_loss: 0.9388 - dense_5_loss: 0.7341 - dense_6_loss: 0.3993 - dense_2_accuracy: 0.6644 - dense_3_accuracy: 0.6500 - dense_4_accuracy: 0.6779 - dense_5_accuracy: 0.7475 - dense_6_accuracy: 0.8682 - val_loss: 2.3838 - val_dense_2_loss: 0.4787 - val_dense_3_loss: 0.5916 - val_dense_4_loss: 0.5908 - val_dense_5_loss: 0.4739 - val_dense_6_loss: 0.2489 - val_dense_2_accuracy: 0.8748 - val_dense_3_accuracy: 0.8451 - val_dense_4_accuracy: 0.8490 - val_dense_5_accuracy: 0.8717 - val_dense_6_accuracy: 0.9328\n",
            "Epoch 3/15\n",
            "60000/60000 [==============================] - 7s 121us/step - loss: 3.5269 - dense_2_loss: 0.8299 - dense_3_loss: 0.8750 - dense_4_loss: 0.8316 - dense_5_loss: 0.6447 - dense_6_loss: 0.3455 - dense_2_accuracy: 0.7209 - dense_3_accuracy: 0.6984 - dense_4_accuracy: 0.7129 - dense_5_accuracy: 0.7781 - dense_6_accuracy: 0.8847 - val_loss: 2.0303 - val_dense_2_loss: 0.3990 - val_dense_3_loss: 0.4945 - val_dense_4_loss: 0.5103 - val_dense_5_loss: 0.4136 - val_dense_6_loss: 0.2128 - val_dense_2_accuracy: 0.8954 - val_dense_3_accuracy: 0.8732 - val_dense_4_accuracy: 0.8749 - val_dense_5_accuracy: 0.8934 - val_dense_6_accuracy: 0.9477\n",
            "Epoch 4/15\n",
            "60000/60000 [==============================] - 7s 117us/step - loss: 3.2137 - dense_2_loss: 0.7464 - dense_3_loss: 0.7932 - dense_4_loss: 0.7630 - dense_5_loss: 0.5950 - dense_6_loss: 0.3162 - dense_2_accuracy: 0.7496 - dense_3_accuracy: 0.7302 - dense_4_accuracy: 0.7363 - dense_5_accuracy: 0.7930 - dense_6_accuracy: 0.8937 - val_loss: 1.7671 - val_dense_2_loss: 0.3420 - val_dense_3_loss: 0.4216 - val_dense_4_loss: 0.4511 - val_dense_5_loss: 0.3638 - val_dense_6_loss: 0.1886 - val_dense_2_accuracy: 0.9112 - val_dense_3_accuracy: 0.8919 - val_dense_4_accuracy: 0.8819 - val_dense_5_accuracy: 0.9071 - val_dense_6_accuracy: 0.9528\n",
            "Epoch 5/15\n",
            "60000/60000 [==============================] - 7s 121us/step - loss: 2.9970 - dense_2_loss: 0.6824 - dense_3_loss: 0.7374 - dense_4_loss: 0.7195 - dense_5_loss: 0.5579 - dense_6_loss: 0.2998 - dense_2_accuracy: 0.7720 - dense_3_accuracy: 0.7486 - dense_4_accuracy: 0.7513 - dense_5_accuracy: 0.8062 - dense_6_accuracy: 0.8980 - val_loss: 1.5877 - val_dense_2_loss: 0.2976 - val_dense_3_loss: 0.3832 - val_dense_4_loss: 0.4031 - val_dense_5_loss: 0.3308 - val_dense_6_loss: 0.1731 - val_dense_2_accuracy: 0.9222 - val_dense_3_accuracy: 0.8965 - val_dense_4_accuracy: 0.8987 - val_dense_5_accuracy: 0.9179 - val_dense_6_accuracy: 0.9585\n",
            "Epoch 6/15\n",
            "60000/60000 [==============================] - 7s 120us/step - loss: 2.8337 - dense_2_loss: 0.6438 - dense_3_loss: 0.6909 - dense_4_loss: 0.6787 - dense_5_loss: 0.5351 - dense_6_loss: 0.2853 - dense_2_accuracy: 0.7861 - dense_3_accuracy: 0.7653 - dense_4_accuracy: 0.7682 - dense_5_accuracy: 0.8123 - dense_6_accuracy: 0.9024 - val_loss: 1.4530 - val_dense_2_loss: 0.2678 - val_dense_3_loss: 0.3444 - val_dense_4_loss: 0.3683 - val_dense_5_loss: 0.3101 - val_dense_6_loss: 0.1624 - val_dense_2_accuracy: 0.9297 - val_dense_3_accuracy: 0.9082 - val_dense_4_accuracy: 0.9069 - val_dense_5_accuracy: 0.9254 - val_dense_6_accuracy: 0.9594\n",
            "Epoch 7/15\n",
            "60000/60000 [==============================] - 7s 117us/step - loss: 2.7000 - dense_2_loss: 0.6070 - dense_3_loss: 0.6575 - dense_4_loss: 0.6495 - dense_5_loss: 0.5111 - dense_6_loss: 0.2750 - dense_2_accuracy: 0.7988 - dense_3_accuracy: 0.7760 - dense_4_accuracy: 0.7775 - dense_5_accuracy: 0.8209 - dense_6_accuracy: 0.9068 - val_loss: 1.3763 - val_dense_2_loss: 0.2618 - val_dense_3_loss: 0.3192 - val_dense_4_loss: 0.3481 - val_dense_5_loss: 0.2922 - val_dense_6_loss: 0.1549 - val_dense_2_accuracy: 0.9298 - val_dense_3_accuracy: 0.9151 - val_dense_4_accuracy: 0.9173 - val_dense_5_accuracy: 0.9295 - val_dense_6_accuracy: 0.9618\n",
            "Epoch 8/15\n",
            "60000/60000 [==============================] - 7s 121us/step - loss: 2.5923 - dense_2_loss: 0.5869 - dense_3_loss: 0.6237 - dense_4_loss: 0.6271 - dense_5_loss: 0.4888 - dense_6_loss: 0.2656 - dense_2_accuracy: 0.8059 - dense_3_accuracy: 0.7907 - dense_4_accuracy: 0.7855 - dense_5_accuracy: 0.8288 - dense_6_accuracy: 0.9086 - val_loss: 1.2732 - val_dense_2_loss: 0.2444 - val_dense_3_loss: 0.2901 - val_dense_4_loss: 0.3274 - val_dense_5_loss: 0.2647 - val_dense_6_loss: 0.1466 - val_dense_2_accuracy: 0.9344 - val_dense_3_accuracy: 0.9218 - val_dense_4_accuracy: 0.9179 - val_dense_5_accuracy: 0.9379 - val_dense_6_accuracy: 0.9652\n",
            "Epoch 9/15\n",
            "60000/60000 [==============================] - 7s 121us/step - loss: 2.5083 - dense_2_loss: 0.5637 - dense_3_loss: 0.6064 - dense_4_loss: 0.6009 - dense_5_loss: 0.4767 - dense_6_loss: 0.2603 - dense_2_accuracy: 0.8127 - dense_3_accuracy: 0.7947 - dense_4_accuracy: 0.7936 - dense_5_accuracy: 0.8337 - dense_6_accuracy: 0.9099 - val_loss: 1.2008 - val_dense_2_loss: 0.2311 - val_dense_3_loss: 0.2712 - val_dense_4_loss: 0.3029 - val_dense_5_loss: 0.2509 - val_dense_6_loss: 0.1447 - val_dense_2_accuracy: 0.9374 - val_dense_3_accuracy: 0.9266 - val_dense_4_accuracy: 0.9261 - val_dense_5_accuracy: 0.9445 - val_dense_6_accuracy: 0.9658\n",
            "Epoch 10/15\n",
            "60000/60000 [==============================] - 7s 122us/step - loss: 2.4403 - dense_2_loss: 0.5520 - dense_3_loss: 0.5889 - dense_4_loss: 0.5816 - dense_5_loss: 0.4650 - dense_6_loss: 0.2528 - dense_2_accuracy: 0.8178 - dense_3_accuracy: 0.8046 - dense_4_accuracy: 0.8012 - dense_5_accuracy: 0.8399 - dense_6_accuracy: 0.9137 - val_loss: 1.1589 - val_dense_2_loss: 0.2245 - val_dense_3_loss: 0.2710 - val_dense_4_loss: 0.2887 - val_dense_5_loss: 0.2359 - val_dense_6_loss: 0.1389 - val_dense_2_accuracy: 0.9390 - val_dense_3_accuracy: 0.9247 - val_dense_4_accuracy: 0.9293 - val_dense_5_accuracy: 0.9421 - val_dense_6_accuracy: 0.9676\n",
            "Epoch 11/15\n",
            "60000/60000 [==============================] - 7s 123us/step - loss: 2.3708 - dense_2_loss: 0.5389 - dense_3_loss: 0.5654 - dense_4_loss: 0.5665 - dense_5_loss: 0.4505 - dense_6_loss: 0.2494 - dense_2_accuracy: 0.8217 - dense_3_accuracy: 0.8101 - dense_4_accuracy: 0.8069 - dense_5_accuracy: 0.8451 - dense_6_accuracy: 0.9152 - val_loss: 1.0913 - val_dense_2_loss: 0.2121 - val_dense_3_loss: 0.2473 - val_dense_4_loss: 0.2732 - val_dense_5_loss: 0.2265 - val_dense_6_loss: 0.1322 - val_dense_2_accuracy: 0.9398 - val_dense_3_accuracy: 0.9310 - val_dense_4_accuracy: 0.9311 - val_dense_5_accuracy: 0.9473 - val_dense_6_accuracy: 0.9694\n",
            "Epoch 12/15\n",
            "60000/60000 [==============================] - 7s 120us/step - loss: 2.3132 - dense_2_loss: 0.5247 - dense_3_loss: 0.5506 - dense_4_loss: 0.5518 - dense_5_loss: 0.4397 - dense_6_loss: 0.2465 - dense_2_accuracy: 0.8277 - dense_3_accuracy: 0.8164 - dense_4_accuracy: 0.8120 - dense_5_accuracy: 0.8476 - dense_6_accuracy: 0.9160 - val_loss: 1.0425 - val_dense_2_loss: 0.2029 - val_dense_3_loss: 0.2412 - val_dense_4_loss: 0.2541 - val_dense_5_loss: 0.2156 - val_dense_6_loss: 0.1287 - val_dense_2_accuracy: 0.9428 - val_dense_3_accuracy: 0.9327 - val_dense_4_accuracy: 0.9337 - val_dense_5_accuracy: 0.9480 - val_dense_6_accuracy: 0.9701\n",
            "Epoch 13/15\n",
            "60000/60000 [==============================] - 7s 118us/step - loss: 2.2644 - dense_2_loss: 0.5107 - dense_3_loss: 0.5419 - dense_4_loss: 0.5381 - dense_5_loss: 0.4331 - dense_6_loss: 0.2407 - dense_2_accuracy: 0.8339 - dense_3_accuracy: 0.8206 - dense_4_accuracy: 0.8177 - dense_5_accuracy: 0.8508 - dense_6_accuracy: 0.9173 - val_loss: 1.0165 - val_dense_2_loss: 0.1984 - val_dense_3_loss: 0.2291 - val_dense_4_loss: 0.2520 - val_dense_5_loss: 0.2088 - val_dense_6_loss: 0.1282 - val_dense_2_accuracy: 0.9440 - val_dense_3_accuracy: 0.9365 - val_dense_4_accuracy: 0.9356 - val_dense_5_accuracy: 0.9506 - val_dense_6_accuracy: 0.9701\n",
            "Epoch 14/15\n",
            "60000/60000 [==============================] - 7s 121us/step - loss: 2.2145 - dense_2_loss: 0.5041 - dense_3_loss: 0.5268 - dense_4_loss: 0.5224 - dense_5_loss: 0.4231 - dense_6_loss: 0.2382 - dense_2_accuracy: 0.8340 - dense_3_accuracy: 0.8242 - dense_4_accuracy: 0.8241 - dense_5_accuracy: 0.8536 - dense_6_accuracy: 0.9179 - val_loss: 0.9992 - val_dense_2_loss: 0.2016 - val_dense_3_loss: 0.2275 - val_dense_4_loss: 0.2404 - val_dense_5_loss: 0.2042 - val_dense_6_loss: 0.1256 - val_dense_2_accuracy: 0.9462 - val_dense_3_accuracy: 0.9382 - val_dense_4_accuracy: 0.9377 - val_dense_5_accuracy: 0.9524 - val_dense_6_accuracy: 0.9717\n",
            "Epoch 15/15\n",
            "60000/60000 [==============================] - 7s 120us/step - loss: 2.1718 - dense_2_loss: 0.4923 - dense_3_loss: 0.5096 - dense_4_loss: 0.5194 - dense_5_loss: 0.4145 - dense_6_loss: 0.2360 - dense_2_accuracy: 0.8388 - dense_3_accuracy: 0.8300 - dense_4_accuracy: 0.8238 - dense_5_accuracy: 0.8567 - dense_6_accuracy: 0.9192 - val_loss: 0.9831 - val_dense_2_loss: 0.1965 - val_dense_3_loss: 0.2165 - val_dense_4_loss: 0.2432 - val_dense_5_loss: 0.1978 - val_dense_6_loss: 0.1291 - val_dense_2_accuracy: 0.9451 - val_dense_3_accuracy: 0.9420 - val_dense_4_accuracy: 0.9392 - val_dense_5_accuracy: 0.9541 - val_dense_6_accuracy: 0.9707\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.callbacks.History at 0x7f0599786748>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CQDF37VaTQcL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Function for calculating Accuracy\n",
        "\n",
        "def calculate_acc(predictions,real_labels):\n",
        "    \n",
        "    individual_counter = 0\n",
        "    global_sequence_counter = 0\n",
        "    for i in range(0,len(predictions[0])):\n",
        "        #Reset sequence counter at the start of each image\n",
        "        sequence_counter = 0 \n",
        "        \n",
        "        for j in range(0,5):\n",
        "            if np.argmax(predictions[j][i]) == np.argmax(real_labels[j][i]):\n",
        "                individual_counter += 1\n",
        "                sequence_counter +=1\n",
        "        \n",
        "        if sequence_counter == 5:\n",
        "            global_sequence_counter += 1\n",
        "         \n",
        "    ind_accuracy = individual_counter/50000.0\n",
        "    global_accuracy = global_sequence_counter/10000.0\n",
        "    \n",
        "    return ind_accuracy,global_accuracy"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eIg6dt2dU9ga",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e20fc053-08ff-4525-f368-1a31243b98c9"
      },
      "source": [
        "predictions = model.predict(test_images)\n",
        "\n",
        "ind_Acc, glob_acc = calculate_acc(predictions, test_labels)\n",
        "print(ind_Acc, glob_acc)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "5.70132 4.6896\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EeRj4bnlVndd",
        "colab_type": "code",
        "outputId": "4d64f52d-bb48-4b15-c11b-fbd86f750579",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 353
        }
      },
      "source": [
        "#Printing some examples of real and predicted labels\n",
        "for i in random.sample(range(0,10000),5):\n",
        "    \n",
        "    actual_labels = []\n",
        "    predicted_labels = []\n",
        "    \n",
        "    for j in range(0,5):\n",
        "        actual_labels.append(np.argmax(test_labels[j][i]))\n",
        "        predicted_labels.append(np.argmax(predictions[j][i]))\n",
        "    print('index', i)\n",
        "    print(\"Actual labels: {}\".format(actual_labels))\n",
        "    print(\"Predicted labels: {}\\n\".format(predicted_labels))"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "index 1058\n",
            "Actual labels: [9, 5, 10, 10, 10]\n",
            "Predicted labels: [9, 5, 10, 10, 10]\n",
            "\n",
            "index 7561\n",
            "Actual labels: [9, 10, 10, 10, 10]\n",
            "Predicted labels: [9, 10, 10, 10, 10]\n",
            "\n",
            "index 5633\n",
            "Actual labels: [9, 3, 9, 5, 1]\n",
            "Predicted labels: [9, 3, 9, 5, 1]\n",
            "\n",
            "index 1878\n",
            "Actual labels: [7, 2, 7, 10, 10]\n",
            "Predicted labels: [7, 2, 7, 10, 10]\n",
            "\n",
            "index 7315\n",
            "Actual labels: [6, 3, 10, 10, 10]\n",
            "Predicted labels: [6, 3, 10, 10, 10]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3SCUxO82WF66",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#A function which return the predicted digit\n",
        "\n",
        "def predict_image(image):\n",
        "  plt.imshow(np.resize(image, (64, 64)), cmap='gray')\n",
        "  plt.show()\n",
        "  image = image.reshape(1, 64, 64, 1)\n",
        "  prediction = model.predict(image)\n",
        "  for i in range(5):\n",
        "    prediction[i] = np.argmax(prediction[i])\n",
        "  a = []\n",
        "  for i in prediction:\n",
        "    if i!=10:\n",
        "      a.append(i)\n",
        "  print('Predicted Value is', ''.join(map(str, a)))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j5tmgtqQj0bA",
        "colab_type": "code",
        "outputId": "ba7fdecd-f08f-4172-be4b-cda0b26a4235",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        }
      },
      "source": [
        "predict_image(test_images[70])"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Predicted Value is 69640\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}