cats_n_dogs_BN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image Classification of Dogs vs. Cats Using CNN Ensemble"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Imports & environment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n",
      "Using gpu device 0: GeForce GTX 980M (CNMeM is enabled with initial size: 90.0% of memory, cuDNN 5105)\n",
      "/home/robert/anaconda3/lib/python3.5/site-packages/theano/sandbox/cuda/__init__.py:600: UserWarning: Your cuDNN version is more recent than the one Theano officially supports. If you see any problems, try updating Theano or downgrading cuDNN to version 5.\n",
      "  warnings.warn(warn)\n",
      "/home/robert/anaconda3/lib/python3.5/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n",
      "  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n",
      "/home/robert/anaconda3/lib/python3.5/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n",
      "  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "\n",
    "from glob import glob\n",
    "from shutil import copyfile\n",
    "from vgg_bn import Vgg16BN\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "\n",
    "ROOT_DIR = os.getcwd()\n",
    "DATA_HOME_DIR = ROOT_DIR + '/data'\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Config & Hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# paths\n",
    "data_path = DATA_HOME_DIR + '/' \n",
    "train_path = data_path + '/train/'\n",
    "valid_path = data_path + '/valid/'\n",
    "test_path = DATA_HOME_DIR + '/test/'\n",
    "model_path = ROOT_DIR + '/models/'\n",
    "submission_path = ROOT_DIR + '/submissions/'\n",
    "\n",
    "# data\n",
    "img_width, img_height = 224, 224\n",
    "batch_size = 64\n",
    "nb_train_samples = 23000\n",
    "nb_valid_samples = 2000\n",
    "nb_test_samples = 12500\n",
    "classes = [\"cats\", \"dogs\"]\n",
    "n_classes = len(classes)\n",
    "\n",
    "# model\n",
    "nb_epoch = 10\n",
    "nb_aug = 5\n",
    "lr = 0.001"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Build the VGG model w/ Batch Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "____________________________________________________________________________________________________\n",
      "Layer (type)                     Output Shape          Param #     Connected to                     \n",
      "====================================================================================================\n",
      "lambda_1 (Lambda)                (None, 3, 224, 224)   0           lambda_input_1[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_1 (ZeroPadding2D)  (None, 3, 226, 226)   0           lambda_1[0][0]                   \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_1 (Convolution2D)  (None, 64, 224, 224)  0           zeropadding2d_1[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_2 (ZeroPadding2D)  (None, 64, 226, 226)  0           convolution2d_1[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_2 (Convolution2D)  (None, 64, 224, 224)  0           zeropadding2d_2[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "maxpooling2d_1 (MaxPooling2D)    (None, 64, 112, 112)  0           convolution2d_2[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_3 (ZeroPadding2D)  (None, 64, 114, 114)  0           maxpooling2d_1[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_3 (Convolution2D)  (None, 128, 112, 112) 0           zeropadding2d_3[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_4 (ZeroPadding2D)  (None, 128, 114, 114) 0           convolution2d_3[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_4 (Convolution2D)  (None, 128, 112, 112) 0           zeropadding2d_4[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "maxpooling2d_2 (MaxPooling2D)    (None, 128, 56, 56)   0           convolution2d_4[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_5 (ZeroPadding2D)  (None, 128, 58, 58)   0           maxpooling2d_2[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_5 (Convolution2D)  (None, 256, 56, 56)   0           zeropadding2d_5[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_6 (ZeroPadding2D)  (None, 256, 58, 58)   0           convolution2d_5[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_6 (Convolution2D)  (None, 256, 56, 56)   0           zeropadding2d_6[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_7 (ZeroPadding2D)  (None, 256, 58, 58)   0           convolution2d_6[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_7 (Convolution2D)  (None, 256, 56, 56)   0           zeropadding2d_7[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "maxpooling2d_3 (MaxPooling2D)    (None, 256, 28, 28)   0           convolution2d_7[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_8 (ZeroPadding2D)  (None, 256, 30, 30)   0           maxpooling2d_3[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_8 (Convolution2D)  (None, 512, 28, 28)   0           zeropadding2d_8[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_9 (ZeroPadding2D)  (None, 512, 30, 30)   0           convolution2d_8[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_9 (Convolution2D)  (None, 512, 28, 28)   0           zeropadding2d_9[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_10 (ZeroPadding2D) (None, 512, 30, 30)   0           convolution2d_9[0][0]            \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_10 (Convolution2D) (None, 512, 28, 28)   0           zeropadding2d_10[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "maxpooling2d_4 (MaxPooling2D)    (None, 512, 14, 14)   0           convolution2d_10[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_11 (ZeroPadding2D) (None, 512, 16, 16)   0           maxpooling2d_4[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_11 (Convolution2D) (None, 512, 14, 14)   0           zeropadding2d_11[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_12 (ZeroPadding2D) (None, 512, 16, 16)   0           convolution2d_11[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_12 (Convolution2D) (None, 512, 14, 14)   0           zeropadding2d_12[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "zeropadding2d_13 (ZeroPadding2D) (None, 512, 16, 16)   0           convolution2d_12[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "convolution2d_13 (Convolution2D) (None, 512, 14, 14)   0           zeropadding2d_13[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "maxpooling2d_5 (MaxPooling2D)    (None, 512, 7, 7)     0           convolution2d_13[0][0]           \n",
      "____________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)              (None, 25088)         0           maxpooling2d_5[0][0]             \n",
      "____________________________________________________________________________________________________\n",
      "dense_1 (Dense)                  (None, 4096)          0           flatten_1[0][0]                  \n",
      "____________________________________________________________________________________________________\n",
      "batchnormalization_1 (BatchNormal(None, 4096)          0           dense_1[0][0]                    \n",
      "____________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)              (None, 4096)          0           batchnormalization_1[0][0]       \n",
      "____________________________________________________________________________________________________\n",
      "dense_2 (Dense)                  (None, 4096)          0           dropout_1[0][0]                  \n",
      "____________________________________________________________________________________________________\n",
      "batchnormalization_2 (BatchNormal(None, 4096)          0           dense_2[0][0]                    \n",
      "____________________________________________________________________________________________________\n",
      "dropout_2 (Dropout)              (None, 4096)          0           batchnormalization_2[0][0]       \n",
      "____________________________________________________________________________________________________\n",
      "dense_4 (Dense)                  (None, 2)             8194        dropout_2[0][0]                  \n",
      "====================================================================================================\n",
      "Total params: 8194\n",
      "____________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "vgg = Vgg16BN(size=(img_width, img_height), n_classes=n_classes, batch_size=batch_size, lr=lr)\n",
    "model = vgg.model\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "info_string = \"{0}x{1}_{2}epoch_{3}aug_{4}lr_vgg16-bn\".format(img_width, img_height, nb_epoch, nb_aug, lr)\n",
    "ckpt_fn = model_path + '{val_loss:.2f}-loss_' + info_string + '.h5'\n",
    "\n",
    "ckpt = ModelCheckpoint(filepath=ckpt_fn,\n",
    "                      monitor='val_loss',\n",
    "                      save_best_only=True,\n",
    "                      save_weights_only=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "vgg.fit(train_path, valid_path,\n",
    "          nb_trn_samples=nb_train_samples,\n",
    "          nb_val_samples=nb_valid_samples,\n",
    "          nb_epoch=nb_epoch,\n",
    "          callbacks=[ckpt],\n",
    "          aug=nb_aug)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predict on Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating predictions for Augmentation... 0\n",
      "Found 12500 images belonging to 1 classes.\n",
      "Generating predictions for Augmentation... 1\n",
      "Found 12500 images belonging to 1 classes.\n",
      "Generating predictions for Augmentation... 2\n",
      "Found 12500 images belonging to 1 classes.\n",
      "Generating predictions for Augmentation... 3\n",
      "Found 12500 images belonging to 1 classes.\n",
      "Generating predictions for Augmentation... 4\n",
      "Found 12500 images belonging to 1 classes.\n",
      "Averaging Predictions Across Augmentations...\n"
     ]
    }
   ],
   "source": [
    "# generate predictions\n",
    "for aug in range(nb_aug):\n",
    "    print(\"Generating predictions for Augmentation {0}...\",format(aug+1))\n",
    "    if aug == 0:\n",
    "        predictions, filenames = vgg.test(test_path, nb_test_samples, aug=nb_aug)\n",
    "    else:\n",
    "        aug_pred, filenames = vgg.test(test_path, nb_test_samples, aug=nb_aug)\n",
    "        predictions += aug_pred\n",
    "\n",
    "print(\"Averaging Predictions Across Augmentations...\")\n",
    "predictions /= nb_aug"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# clip predictions\n",
    "c = 0.01\n",
    "preds = np.clip(predictions, c, 1-c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Writing Predictions to CSV...\n",
      "0 / 12500\n",
      "2500 / 12500\n",
      "5000 / 12500\n",
      "7500 / 12500\n",
      "10000 / 12500\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "sub_file = submission_path + info_string + '.csv'\n",
    "\n",
    "with open(sub_file, 'w') as f:\n",
    "    print(\"Writing Predictions to CSV...\")\n",
    "    f.write('id,label\\n')\n",
    "    for i, image_name in enumerate(filenames):\n",
    "        pred = ['%.6f' % p for p in preds[i, :]]\n",
    "        if i % 2500 == 0:\n",
    "            print(i, '/', nb_test_samples)\n",
    "        f.write('%s,%s\\n' % (os.path.basename(image_name).replace('.jpg', ''), (pred[1])))\n",
    "    print(\"Done.\")"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "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": 1
}