Indoor vs Outdoor.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import np_utils\n",
    "import numpy as np\n",
    "import keras\n",
    "from time import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "datagen=keras.preprocessing.image.ImageDataGenerator(rescale=1./255, zoom_range=0.2, horizontal_flip=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1979 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "\n",
    "train_data = datagen.flow_from_directory(\n",
    "        directory=r'/home/prajacta/yt8m/youtube-8m-videos-frames/train1',\n",
    "        target_size=(64, 64),\n",
    "        batch_size=32,\n",
    "        class_mode=\"binary\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 250 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "test_data = datagen.flow_from_directory(\n",
    "    directory=r\"/home/prajacta/yt8m/youtube-8m-videos-frames/1\",\n",
    "    target_size=(64, 64),\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=32,\n",
    "    class_mode=\"binary\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 246 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "valid_data = datagen.flow_from_directory(\n",
    "    directory=r\"/home/prajacta/yt8m/youtube-8m-videos-frames/val1\",\n",
    "    target_size=(64, 64),\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=32,\n",
    "    class_mode=\"binary\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 62, 62, 32)        896       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 62, 62, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 31, 31, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 29, 29, 32)        9248      \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 29, 29, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 12, 12, 64)        18496     \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 12, 12, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 6, 6, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 2304)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                147520    \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 65        \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 1)                 0         \n",
      "=================================================================\n",
      "Total params: 176,225\n",
      "Trainable params: 176,225\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "from keras.layers import Activation, Dropout, Flatten, Dense\n",
    "#Adding sequential model in keras\n",
    "model = Sequential()\n",
    "# Image size 64*64 \n",
    "model.add(Conv2D(32, (3, 3), input_shape=(64, 64, 3)))\n",
    "#ReLu adds nonlinearity to the network.\n",
    "model.add(Activation('relu'))\n",
    "# Pooling helps to reduce dimensions.\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Conv2D(32, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Conv2D(64, (3, 3)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "#Fully connected layer to detect final high level features.\n",
    "model.add(Flatten())  \n",
    "model.add(Dense(64))\n",
    "model.add(Activation('relu'))\n",
    "#Dropout is added to reduce overfitting.\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(Dense(1))\n",
    "model.add(Activation('sigmoid'))\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer='rmsprop',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "1979/1979 [==============================] - 732s 370ms/step - loss: 0.4304 - acc: 0.7968 - val_loss: 0.4717 - val_acc: 0.8252\n",
      "Epoch 2/50\n",
      "1979/1979 [==============================] - 748s 378ms/step - loss: 0.2218 - acc: 0.9126 - val_loss: 0.4829 - val_acc: 0.8049\n",
      "Epoch 3/50\n",
      "1979/1979 [==============================] - 754s 381ms/step - loss: 0.1797 - acc: 0.9351 - val_loss: 0.7477 - val_acc: 0.8171\n",
      "Epoch 4/50\n",
      "1979/1979 [==============================] - 749s 379ms/step - loss: 0.1868 - acc: 0.9363 - val_loss: 0.8340 - val_acc: 0.8171\n",
      "Epoch 5/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.1990 - acc: 0.9338 - val_loss: 0.5731 - val_acc: 0.7764\n",
      "Epoch 6/50\n",
      "1979/1979 [==============================] - 757s 383ms/step - loss: 0.2157 - acc: 0.9299 - val_loss: 0.7157 - val_acc: 0.8008\n",
      "Epoch 7/50\n",
      "1979/1979 [==============================] - 750s 379ms/step - loss: 0.2172 - acc: 0.9284 - val_loss: 1.1912 - val_acc: 0.8211\n",
      "Epoch 8/50\n",
      "1979/1979 [==============================] - 756s 382ms/step - loss: 0.2378 - acc: 0.9245 - val_loss: 1.1830 - val_acc: 0.8333\n",
      "Epoch 9/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.2222 - acc: 0.9286 - val_loss: 0.6972 - val_acc: 0.8496\n",
      "Epoch 10/50\n",
      "1979/1979 [==============================] - 738s 373ms/step - loss: 0.2294 - acc: 0.9278 - val_loss: 0.6715 - val_acc: 0.8699\n",
      "Epoch 11/50\n",
      "1979/1979 [==============================] - 754s 381ms/step - loss: 0.2055 - acc: 0.9326 - val_loss: 1.1466 - val_acc: 0.8008\n",
      "Epoch 12/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.2087 - acc: 0.9341 - val_loss: 0.6105 - val_acc: 0.8699\n",
      "Epoch 13/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.2059 - acc: 0.9346 - val_loss: 0.4687 - val_acc: 0.8455\n",
      "Epoch 14/50\n",
      "1979/1979 [==============================] - 728s 368ms/step - loss: 0.2174 - acc: 0.9336 - val_loss: 1.0434 - val_acc: 0.8333\n",
      "Epoch 15/50\n",
      "1979/1979 [==============================] - 729s 368ms/step - loss: 0.2371 - acc: 0.9340 - val_loss: 0.7679 - val_acc: 0.8293\n",
      "Epoch 16/50\n",
      "1979/1979 [==============================] - 728s 368ms/step - loss: 0.2252 - acc: 0.9367 - val_loss: 0.4537 - val_acc: 0.7967\n",
      "Epoch 17/50\n",
      "1979/1979 [==============================] - 728s 368ms/step - loss: 0.1973 - acc: 0.9366 - val_loss: 0.6708 - val_acc: 0.8293\n",
      "Epoch 18/50\n",
      "1979/1979 [==============================] - 729s 368ms/step - loss: 0.2165 - acc: 0.9349 - val_loss: 0.7978 - val_acc: 0.8171\n",
      "Epoch 19/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.2230 - acc: 0.9354 - val_loss: 0.6843 - val_acc: 0.7927\n",
      "Epoch 20/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.2420 - acc: 0.9321 - val_loss: 0.7895 - val_acc: 0.7805\n",
      "Epoch 21/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.2357 - acc: 0.9322 - val_loss: 1.1301 - val_acc: 0.8130\n",
      "Epoch 22/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.2323 - acc: 0.9310 - val_loss: 0.5995 - val_acc: 0.8211\n",
      "Epoch 23/50\n",
      "1979/1979 [==============================] - 720s 364ms/step - loss: 0.2490 - acc: 0.9276 - val_loss: 0.7040 - val_acc: 0.8496\n",
      "Epoch 24/50\n",
      "1979/1979 [==============================] - 710s 359ms/step - loss: 0.2662 - acc: 0.9302 - val_loss: 0.9392 - val_acc: 0.8659\n",
      "Epoch 25/50\n",
      "1979/1979 [==============================] - 711s 359ms/step - loss: 0.2503 - acc: 0.9291 - val_loss: 0.9246 - val_acc: 0.8211\n",
      "Epoch 26/50\n",
      "1979/1979 [==============================] - 711s 359ms/step - loss: 0.2640 - acc: 0.9252 - val_loss: 1.5680 - val_acc: 0.7886\n",
      "Epoch 27/50\n",
      "1979/1979 [==============================] - 710s 359ms/step - loss: 0.2936 - acc: 0.9220 - val_loss: 1.1181 - val_acc: 0.8333\n",
      "Epoch 28/50\n",
      "1979/1979 [==============================] - 710s 359ms/step - loss: 0.2746 - acc: 0.9210 - val_loss: 1.4169 - val_acc: 0.7642\n",
      "Epoch 29/50\n",
      "1979/1979 [==============================] - 709s 358ms/step - loss: 0.2885 - acc: 0.9198 - val_loss: 0.8181 - val_acc: 0.8293\n",
      "Epoch 30/50\n",
      "1979/1979 [==============================] - 720s 364ms/step - loss: 0.4495 - acc: 0.9067 - val_loss: 6.0672 - val_acc: 0.6138\n",
      "Epoch 31/50\n",
      "1979/1979 [==============================] - 734s 371ms/step - loss: 1.6217 - acc: 0.8457 - val_loss: 0.7082 - val_acc: 0.8171\n",
      "Epoch 32/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.4033 - acc: 0.9119 - val_loss: 1.0217 - val_acc: 0.7480\n",
      "Epoch 33/50\n",
      "1979/1979 [==============================] - 732s 370ms/step - loss: 0.3138 - acc: 0.9100 - val_loss: 0.7968 - val_acc: 0.8008\n",
      "Epoch 34/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.3241 - acc: 0.9129 - val_loss: 0.6516 - val_acc: 0.7886\n",
      "Epoch 35/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.3068 - acc: 0.9119 - val_loss: 0.8352 - val_acc: 0.8333\n",
      "Epoch 36/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.3370 - acc: 0.9046 - val_loss: 0.8015 - val_acc: 0.8211\n",
      "Epoch 37/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.3383 - acc: 0.9091 - val_loss: 0.6796 - val_acc: 0.8008\n",
      "Epoch 38/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.3290 - acc: 0.9108 - val_loss: 0.7908 - val_acc: 0.8089\n",
      "Epoch 39/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.3470 - acc: 0.9077 - val_loss: 1.3308 - val_acc: 0.8211\n",
      "Epoch 40/50\n",
      "1979/1979 [==============================] - 730s 369ms/step - loss: 0.3155 - acc: 0.9111 - val_loss: 1.7110 - val_acc: 0.8130\n",
      "Epoch 41/50\n",
      "1979/1979 [==============================] - 728s 368ms/step - loss: 0.3429 - acc: 0.9071 - val_loss: 0.8930 - val_acc: 0.8211\n",
      "Epoch 42/50\n",
      "1979/1979 [==============================] - 729s 368ms/step - loss: 0.4646 - acc: 0.9029 - val_loss: 1.0506 - val_acc: 0.7846\n",
      "Epoch 43/50\n",
      "1979/1979 [==============================] - 731s 369ms/step - loss: 0.3506 - acc: 0.9102 - val_loss: 0.8480 - val_acc: 0.7967\n",
      "Epoch 44/50\n",
      "1979/1979 [==============================] - 732s 370ms/step - loss: 0.3518 - acc: 0.9056 - val_loss: 1.4190 - val_acc: 0.7764\n",
      "Epoch 45/50\n",
      "1979/1979 [==============================] - 733s 370ms/step - loss: 0.3830 - acc: 0.9042 - val_loss: 0.8542 - val_acc: 0.8089\n",
      "Epoch 46/50\n",
      "1979/1979 [==============================] - 733s 370ms/step - loss: 0.3562 - acc: 0.9065 - val_loss: 0.8824 - val_acc: 0.7642\n",
      "Epoch 47/50\n",
      "1979/1979 [==============================] - 732s 370ms/step - loss: 0.6733 - acc: 0.8848 - val_loss: 0.9409 - val_acc: 0.7561\n",
      "Epoch 48/50\n",
      "1979/1979 [==============================] - 731s 370ms/step - loss: 0.3668 - acc: 0.9028 - val_loss: 1.4821 - val_acc: 0.7683\n",
      "Epoch 49/50\n",
      "1979/1979 [==============================] - 731s 370ms/step - loss: 0.3903 - acc: 0.8940 - val_loss: 1.6761 - val_acc: 0.7683\n",
      "Epoch 50/50\n",
      "1979/1979 [==============================] - 731s 370ms/step - loss: 1.4130 - acc: 0.8397 - val_loss: 0.9199 - val_acc: 0.8171\n"
     ]
    }
   ],
   "source": [
    "cp = keras.callbacks.ModelCheckpoint('/home/prajacta/yt8m', \n",
    "                                                 save_weights_only=True,\n",
    "                                                 verbose=1)\n",
    "tb=keras.callbacks.TensorBoard(log_dir='./home'.format(time()), histogram_freq=0,  \n",
    "          write_graph=True, write_images=True)\n",
    "from PIL import ImageFile\n",
    "ImageFile.LOAD_TRUNCATED_IMAGES = True\n",
    "history=model.fit_generator(\n",
    "        train_data,\n",
    "        steps_per_epoch=train_data.n,\n",
    "        epochs=50,\n",
    "        validation_data=valid_data,\n",
    "        validation_steps=valid_data.n, callbacks = [tb])\n",
    "model.save('mymodel.h5') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1.128835253715515, 0.7959999980926513]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate_generator(test_data, len(test_data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "probabilities = model.predict_generator(test_data, len(test_data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[60, 65],\n",
       "       [66, 59]])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "y_true = np.array([0] * 125 + [1] * 125)\n",
    "y_pred = probabilities > 0.5\n",
    "\n",
    "confusion_matrix(y_true, y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1]]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from keras.preprocessing import image\n",
    "img = image.load_img('test3.jpg', target_size=(64, 64))\n",
    "x = image.img_to_array(img)\n",
    "x = np.expand_dims(x, axis=0)\n",
    "\n",
    "images = np.vstack([x])\n",
    "classes = model.predict_classes(images, batch_size=10)\n",
    "print classes\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "label=train_data.class_indices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'indoor': 0, 'outdoor': 1}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "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"
    },
    {
     "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": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "# plots keras\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Plot training & validation loss values\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Model loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import plot_model\n",
    "plot_model(model, to_file='model.png')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}