TTA.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Z6_pnIpaQ6LC",
    "outputId": "4ab515ec-9903-4ab8-fa99-8aefc68f30f3"
   },
   "outputs": [],
   "source": [
    "from keras.datasets import cifar10, mnist, fashion_mnist\n",
    "from keras.preprocessing import image\n",
    "from keras.layers import Conv2D, Dense, Input, Flatten, Dropout\n",
    "from keras.models import Model, Sequential, load_model\n",
    "from keras.utils import to_categorical\n",
    "from keras import optimizers\n",
    "import keras\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "from keras import backend as K\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VZ-2tCPzSl6Z"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
      "170500096/170498071 [==============================] - 831s 5us/step\n"
     ]
    }
   ],
   "source": [
    "(x_train, y_train), (x_val, y_val) = cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hQWOHFP6Ss63"
   },
   "outputs": [],
   "source": [
    "x_train, x_val = x_train.astype('float32')/255.,  x_val.astype('float32')/255.\n",
    "y_train, y_val = to_categorical(y_train), to_categorical(y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "46C8mG0Nwlcj"
   },
   "outputs": [],
   "source": [
    "train_datagen = ImageDataGenerator(\n",
    "        shear_range=0.1,\n",
    "        zoom_range=0.1,\n",
    "        horizontal_flip=True,\n",
    "        rotation_range=10.,\n",
    "        fill_mode='reflect',\n",
    "        width_shift_range = 0.1, \n",
    "        height_shift_range = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DFeId4b_wxEg"
   },
   "outputs": [],
   "source": [
    "train_datagen.fit(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 388
    },
    "colab_type": "code",
    "id": "D0h-cpVlQ6LX",
    "outputId": "69c63b5c-2bbb-4394-8e8f-afabe1669fd2"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 15 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,5))\n",
    "\n",
    "for X_batch, y_batch in train_datagen.flow(x_train, y_train, batch_size=15):\n",
    "    for i in range(0, 15):\n",
    "        plt.subplot(3, 5, i+1)\n",
    "        plt.imshow(X_batch[i].reshape(32, 32, 3), cmap=plt.get_cmap('gray'))\n",
    "        plt.axis('off')\n",
    "    plt.tight_layout(pad=0.1)\n",
    "    plt.show()\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 139
    },
    "colab_type": "code",
    "id": "dMttrysMQ6Lc",
    "outputId": "5c2d1805-886b-47fc-cc49-b760307b82fe"
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Conv2D(64,(3,3), activation='relu', input_shape=(32,32,3)))\n",
    "model.add(Conv2D(128,(3,3), activation='relu'))\n",
    "model.add(Conv2D(128,(3,3), activation='relu'))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(128, activation='relu'))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vEZ8kGoNQ6Lf"
   },
   "outputs": [],
   "source": [
    "bs = 64\n",
    "epochs = 15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 598
    },
    "colab_type": "code",
    "id": "zuFhL-uKxXap",
    "outputId": "386213cd-9f04-46e1-86f5-0a2d595a398e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/cloudy/anaconda3/envs/compu/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:422: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n",
      "\n",
      "Epoch 1/15\n",
      "782/781 [==============================] - 61s 79ms/step - loss: 1.6536 - accuracy: 0.3958 - val_loss: 1.3026 - val_accuracy: 0.5317\n",
      "Epoch 2/15\n",
      "782/781 [==============================] - 60s 77ms/step - loss: 1.3096 - accuracy: 0.5334 - val_loss: 1.0993 - val_accuracy: 0.6068\n",
      "Epoch 3/15\n",
      "782/781 [==============================] - 76s 97ms/step - loss: 1.1552 - accuracy: 0.5913 - val_loss: 0.9430 - val_accuracy: 0.6702\n",
      "Epoch 4/15\n",
      "782/781 [==============================] - 95s 121ms/step - loss: 1.0732 - accuracy: 0.6231 - val_loss: 0.9204 - val_accuracy: 0.6795\n",
      "Epoch 5/15\n",
      "782/781 [==============================] - 103s 132ms/step - loss: 1.0117 - accuracy: 0.6466 - val_loss: 0.8561 - val_accuracy: 0.7074\n",
      "Epoch 6/15\n",
      "782/781 [==============================] - 110s 141ms/step - loss: 0.9617 - accuracy: 0.6624 - val_loss: 0.8737 - val_accuracy: 0.6952\n",
      "Epoch 7/15\n",
      "782/781 [==============================] - 115s 147ms/step - loss: 0.9321 - accuracy: 0.6759 - val_loss: 0.8138 - val_accuracy: 0.7161\n",
      "Epoch 8/15\n",
      "782/781 [==============================] - 119s 152ms/step - loss: 0.9038 - accuracy: 0.6866 - val_loss: 0.7579 - val_accuracy: 0.7375\n",
      "Epoch 9/15\n",
      "782/781 [==============================] - 123s 157ms/step - loss: 0.8816 - accuracy: 0.6920 - val_loss: 0.7353 - val_accuracy: 0.7448\n",
      "Epoch 10/15\n",
      "782/781 [==============================] - 122s 156ms/step - loss: 0.8576 - accuracy: 0.7015 - val_loss: 0.7926 - val_accuracy: 0.7286\n",
      "Epoch 11/15\n",
      "782/781 [==============================] - 128s 163ms/step - loss: 0.8411 - accuracy: 0.7062 - val_loss: 0.7560 - val_accuracy: 0.7390\n",
      "Epoch 12/15\n",
      "782/781 [==============================] - 128s 164ms/step - loss: 0.8218 - accuracy: 0.7170 - val_loss: 0.7221 - val_accuracy: 0.7507\n",
      "Epoch 13/15\n",
      "782/781 [==============================] - 135s 172ms/step - loss: 0.8090 - accuracy: 0.7186 - val_loss: 0.7335 - val_accuracy: 0.7535\n",
      "Epoch 14/15\n",
      "782/781 [==============================] - 135s 173ms/step - loss: 0.7965 - accuracy: 0.7250 - val_loss: 0.7195 - val_accuracy: 0.7550\n",
      "Epoch 15/15\n",
      "782/781 [==============================] - 132s 168ms/step - loss: 0.7856 - accuracy: 0.7264 - val_loss: 0.6921 - val_accuracy: 0.7650\n"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(train_datagen.flow(x_train, y_train,\n",
    "                              batch_size=bs),\n",
    "                              epochs=epochs,\n",
    "                              steps_per_epoch=len(x_train)/bs,\n",
    "                              validation_data=(x_val, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "5XAeIaCJQ6Lp",
    "outputId": "2f9e2fe7-1eb3-4f48-cac9-24baceb09af9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 8s 776us/step\n",
      "Baseline accuracy: 0.7649999856948853\n"
     ]
    }
   ],
   "source": [
    "print(f'Baseline accuracy: {model.evaluate(x_val, y_val)[1]}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_ICzIOrPQ6Lu"
   },
   "source": [
    "### Define the test Image Generator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "9zJGPFtEQ6Lw"
   },
   "outputs": [],
   "source": [
    "tta_steps = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ZVU-2IQQQ6Lz"
   },
   "outputs": [],
   "source": [
    "test_datagen = ImageDataGenerator(\n",
    "        shear_range=0.1,\n",
    "        zoom_range=0.1,\n",
    "        horizontal_flip=True,\n",
    "        rotation_range=10.,\n",
    "        fill_mode='reflect', \n",
    "        width_shift_range = 0.1, \n",
    "        height_shift_range = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "-kgjEFe-Q6L3",
    "outputId": "d64e4412-5108-4b29-ffd0-37dd0e1e0a9d"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [01:18<00:00,  7.87s/it]\n"
     ]
    }
   ],
   "source": [
    "predictions = []\n",
    "\n",
    "for i in tqdm(range(tta_steps)):\n",
    "    preds = model.predict_generator(test_datagen.flow(x_val, batch_size=bs, shuffle=False), steps = len(x_val)/bs)\n",
    "    predictions.append(preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "l61pZL8HQ6L7"
   },
   "outputs": [],
   "source": [
    "final_pred = np.mean(predictions, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "DmmBkyvoQ6MD",
    "outputId": "91ea2744-ef11-4452-e01f-674672ac17fc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy with TTA: 0.7834\n"
     ]
    }
   ],
   "source": [
    "print(f'Accuracy with TTA: {np.mean(np.equal(np.argmax(y_val, axis=-1), np.argmax(final_pred, axis=-1)))}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "m1dcQj2-Q6MJ"
   },
   "source": [
    "### See the impact of TTA on a single image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 362
    },
    "colab_type": "code",
    "id": "cEA8OgcBQ6MJ",
    "outputId": "e0755773-907b-4382-c08f-942d5de6522b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "test_image = x_val[9]\n",
    "plt.imshow(test_image)\n",
    "plt.title(f'real label: {np.argmax(y_val[image_ix])}')\n",
    "plt.axis('off');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 255
    },
    "colab_type": "code",
    "id": "3Zt-DoKJQ6MP",
    "outputId": "3f8b9114-41a3-4d67-e15b-e1d7383844e3"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1008 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20,14))\n",
    "predictions = []\n",
    "\n",
    "for i in range(5):\n",
    "    im = test_datagen.flow(test_image.reshape(1,32,32,3), batch_size=1, shuffle=False)\n",
    "    im = next(im)\n",
    "    pred = model.predict(im.reshape(1,32,32,3))\n",
    "    predictions.append(pred)\n",
    "    plt.subplot(1, 5, i+1)\n",
    "    plt.title(f'pred label: {np.argmax(pred)}')\n",
    "    plt.imshow(im.reshape(32, 32, 3), cmap=plt.get_cmap('gray'))\n",
    "    plt.axis('off')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Ozb5T7DqQ6MV",
    "outputId": "64a97335-a864-4cdf-ec9c-205bfb1cdac6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "final pred: 1\n"
     ]
    }
   ],
   "source": [
    "final_pred = np.mean(predictions, axis=0)\n",
    "print(f'final pred: {np.argmax(final_pred)}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "TTA.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}