dog_breed_classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hands on session for image classification. Here a data set from kaggle for dog breed identification challenge is being used for illustration, however this same workflow can be translated into other image classification problems as well.\n",
    "Download the data from here: https://www.kaggle.com/c/dog-breed-identification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setting the Environment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Install Anaconda Python 3 version: https://www.anaconda.com/download/\n",
    "Once the installation finishes, you will have to create a virtual environment called conda environment with the python tools required for your general ML development, so that it does not interfere with your bare metal python installations and avoid other version mismatch issues. To create a conda environment, follow these instructions\n",
    "1. The default installation for python 3 in anaconda is 3.7 but tensorflow has not yet released a compatible version for 3.7\n",
    "2. So we are going to create a vitrual environment in python 3.4/3.5/3.6 Currently tensorflow\n",
    "3. If you are using GPU you should check for compatible CUDA and CuDNN libraries.\n",
    "4. If you are using windows go to start and open anaconda prompt, if you are using UNIX based system add conda to your path\n",
    "5. Run these commands: \n",
    "        >> conda create -n dog_breed_env python=3.5\n",
    "        >> conda activate dog_breed_env\n",
    "        >> pip install tensorflow-gpu\n",
    "        >> pip install pandas\n",
    "        >> pip install keras\n",
    "        >> pip install jupyterlab\n",
    "6. For more Details refer: https://conda.io/docs/user-guide/tasks/manage-environments.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import keras\n",
    "from keras.layers import Dense, Activation, Conv2D,MaxPooling2D,Flatten, Dropout\n",
    "from keras.models import Sequential\n",
    "import os\n",
    "import sys\n",
    "import cv2\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Read the Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "base_dir=os.getcwd()\n",
    "data=pd.read_csv(\"labels.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inspect the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>breed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>000bec180eb18c7604dcecc8fe0dba07</td>\n",
       "      <td>boston_bull</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>001513dfcb2ffafc82cccf4d8bbaba97</td>\n",
       "      <td>dingo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>001cdf01b096e06d78e9e5112d419397</td>\n",
       "      <td>pekinese</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00214f311d5d2247d5dfe4fe24b2303d</td>\n",
       "      <td>bluetick</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0021f9ceb3235effd7fcde7f7538ed62</td>\n",
       "      <td>golden_retriever</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 id             breed\n",
       "0  000bec180eb18c7604dcecc8fe0dba07       boston_bull\n",
       "1  001513dfcb2ffafc82cccf4d8bbaba97             dingo\n",
       "2  001cdf01b096e06d78e9e5112d419397          pekinese\n",
       "3  00214f311d5d2247d5dfe4fe24b2303d          bluetick\n",
       "4  0021f9ceb3235effd7fcde7f7538ed62  golden_retriever"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1dcfdf5c748>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "img_filepath=os.path.join(base_dir,'train',data['id'][4]+'.jpg')\n",
    "#print(img_filepath)\n",
    "import matplotlib.image as mpimg\n",
    "img=mpimg.imread(img_filepath)\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number of classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "classes=data['breed'].unique()\n",
    "len(classes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "One hot encoding classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "oh_array=np.ones([len(classes)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "one_hot_dictionary={}\n",
    "class_count=0\n",
    "for c in classes:\n",
    "    oh_array=np.zeros([len(classes)])\n",
    "    oh_array[class_count]=1\n",
    "    one_hot_dictionary[c]=oh_array\n",
    "    class_count+=1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Writing a custom generator, there are image generators available in keras that you can use, please go through the documentation here: https://keras.io/preprocessing/image/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def doggy_generator(start=0,stop=7000,reshape_dim=[512,512]):\n",
    "    while True:\n",
    "        for a in range(start,stop):\n",
    "            img_filepath=os.path.join(base_dir,'train',data['id'][a]+'.jpg')\n",
    "            #print(img_filepath)\n",
    "            img=cv2.imread(img_filepath)\n",
    "            resized_img=cv2.resize(img,(reshape_dim[0],reshape_dim[1]),interpolation=cv2.INTER_CUBIC)\n",
    "            norm_image=cv2.normalize(resized_img, None, alpha=0, beta=1, norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_32F)\n",
    "            img_np=np.zeros([1,reshape_dim[0],reshape_dim[1],3])\n",
    "            breed=data['breed'][a]\n",
    "            oh_label=one_hot_dictionary[breed]\n",
    "            label_np=np.zeros([1,120])\n",
    "            img_np[0,:,:,:]=norm_image\n",
    "            label_np[0,:]=oh_label\n",
    "            yield img_np,label_np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "traingen=doggy_generator()\n",
    "testgen=doggy_generator(7000,len(data['id']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "a=traingen.__next__()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 120)"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a[1].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def dog_breed_classifier_model(img_row,img_col,classes): \n",
    "    model=Sequential()\n",
    "\n",
    "    model.add(Conv2D(32,3,3,input_shape=(img_row,img_col,3)))\n",
    "    model.add(Activation('relu'))\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(Conv2D(64,3,3))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "    model.add(Conv2D(64,4,4))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(Conv2D(64,3,3))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "    model.add(Conv2D(64,4,4))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(Conv2D(64,3,3))\n",
    "    model.add(Activation('relu'))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(256, activation='relu'))\n",
    "    model.add(Dense(256,activation='relu'))\n",
    "    model.add(Dropout(0.5))\n",
    "    model.add(Dense(256,activation='relu'))\n",
    "    model.add(Dropout(0.5))\n",
    "    model.add(Dense(classes))\n",
    "    model.add(Activation('softmax'))\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:7: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(32, (3, 3), input_shape=(512, 512,...)`\n",
      "  import sys\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:9: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(32, (3, 3))`\n",
      "  if __name__ == '__main__':\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:13: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3))`\n",
      "  del sys.path[0]\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:15: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3))`\n",
      "  from ipykernel import kernelapp as app\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:19: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (4, 4))`\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:21: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3))`\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:25: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (4, 4))`\n",
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:27: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3))`\n"
     ]
    }
   ],
   "source": [
    "model=dog_breed_classifier_model(512,512,120)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.compile(loss='categorical_crossentropy',optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Please do not run this in the hands on session, it takes few hours to optimize\n",
    "model.fit_generator(traingen,steps_per_epoch=7000,epochs=10,validation_data=testgen,validation_steps=3000)\n",
    "tl_model.save('dog_breed_classifier_model_filepath.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transfer Learning\n",
    "When we use the weights or architecture of the models which were trained earlier for similar cases, it's called transfer learning. We can use the same for our case. Many models were trained on natural images for the Imagenet challenge. These models are highly complex and have been trained on 100s of GPUs optimized for much larger classification problem."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.applications.inception_v3 import InceptionV3\n",
    "from keras.layers import GlobalAveragePooling2D\n",
    "from keras.models import Model\n",
    "from keras.optimizers import SGD\n",
    "\n",
    "def add_new_last_layer(base_model,classes):    \n",
    "    '''\n",
    "    Function to add new trainable layers to the base models\n",
    "    '''\n",
    "    x=base_model.output\n",
    "    x=GlobalAveragePooling2D()(x)\n",
    "    x=Dense(1024, activation='relu')(x)\n",
    "    predict=Dense(classes, activation='softmax')(x)\n",
    "    model = Model(input=base_model.input, output=predict)\n",
    "    return model\n",
    "\n",
    "def setup_finetuning(model,freeze_layers):\n",
    "    '''\n",
    "    Function to freeze layers in the model and return compiled model\n",
    "    '''\n",
    "    for layer in model.layers[:freeze_layers]:\n",
    "        layer.trainable=False\n",
    "    for layer in model.layers[freeze_layers:]:\n",
    "        layer.trainable= True\n",
    "    model.compile(optimizer=SGD(lr=0.0001, momentum=0.9),loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "    \n",
    "def transfer_learning_setup(model, basemodel):\n",
    "    '''\n",
    "    Another function to freeze base model layers and return compiled model\n",
    "    '''\n",
    "    for layer in basemodel.layers:\n",
    "        layer.trainable=False\n",
    "    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Inception model with final layer changed\n",
    "base_model = InceptionV3(weights='imagenet', include_top=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\anand\\anaconda3\\envs\\py35\\lib\\site-packages\\ipykernel_launcher.py:14: UserWarning: Update your `Model` call to the Keras 2 API: `Model(outputs=Tensor(\"de..., inputs=Tensor(\"in...)`\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "tl_model=add_new_last_layer(base_model,120)\n",
    "setup_finetuning(tl_model,172)\n",
    "#tl_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.callbacks import ModelCheckpoint\n",
    "checkpointer = ModelCheckpoint(filepath='dog_breed_classifier.h5', save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "7000/7000 [==============================] - 699s 100ms/step - loss: 4.8053 - acc: 0.0107 - val_loss: 15.6240 - val_acc: 0.0070\n",
      "Epoch 2/10\n",
      "7000/7000 [==============================] - 692s 99ms/step - loss: 4.7797 - acc: 0.0117 - val_loss: 15.5782 - val_acc: 0.0063\n",
      "Epoch 3/10\n",
      "7000/7000 [==============================] - 694s 99ms/step - loss: 4.7650 - acc: 0.0153 - val_loss: 15.6258 - val_acc: 0.0080\n",
      "Epoch 4/10\n",
      "7000/7000 [==============================] - 694s 99ms/step - loss: 4.7465 - acc: 0.0170 - val_loss: 15.6404 - val_acc: 0.0073\n",
      "Epoch 5/10\n",
      "7000/7000 [==============================] - 691s 99ms/step - loss: 4.7032 - acc: 0.0220 - val_loss: 15.8060 - val_acc: 0.0107\n",
      "Epoch 6/10\n",
      "7000/7000 [==============================] - 690s 99ms/step - loss: 4.5362 - acc: 0.0280 - val_loss: 15.7740 - val_acc: 0.0117\n",
      "Epoch 7/10\n",
      "7000/7000 [==============================] - 691s 99ms/step - loss: 4.2636 - acc: 0.0324 - val_loss: 15.7277 - val_acc: 0.0117\n",
      "Epoch 8/10\n",
      "7000/7000 [==============================] - 691s 99ms/step - loss: 4.0454 - acc: 0.0436 - val_loss: 15.7615 - val_acc: 0.0063\n",
      "Epoch 9/10\n",
      "7000/7000 [==============================] - 690s 99ms/step - loss: 3.8662 - acc: 0.0529 - val_loss: 15.7528 - val_acc: 0.0083\n",
      "Epoch 10/10\n",
      "7000/7000 [==============================] - 690s 99ms/step - loss: 3.7327 - acc: 0.0657 - val_loss: 15.9677 - val_acc: 0.0060\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1de6ef88f60>"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tl_model.fit_generator(traingen,steps_per_epoch=7000,epochs=100,validation_data=testgen,validation_steps=3000,callbacks=[checkpointer])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "'''\n",
    "HW- Write a Prediction function\n",
    "def prediction_function(test_img_filepath,model):\n",
    "    img=cv2.imread(test_img_filepath)\n",
    "    resized_img=cv2.resize(img,(reshape_dim[0],reshape_dim[1]),interpolation=cv2.INTER_CUBIC)\n",
    "    norm_image=cv2.normalize(resized_img, None, alpha=0, beta=1, norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_32F)\n",
    "    img_np=np.zeros([1,512,512,3])\n",
    "    img_np[0,:,:,:]=norm_img\n",
    "    predict_one_hot-model.predict(img_np)\n",
    "    \"\"\"\n",
    "    You will get an array of (1,120) elements similar to the label_np in our generator function\n",
    "    Fill out the rest of the function here to extract and return the highest probable class and it's probability.\n",
    "    \"\"\"\n",
    "'''"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use Pretrained Models to Predict\n",
    "We will be using resnet50 model which was trained on imagenet challenge data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels.h5\n",
      "102858752/102853048 [==============================] - 19s 0us/step\n",
      "Downloading data from https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json\n",
      "40960/35363 [==================================] - 0s 1us/step\n",
      "Predicted: [('n02099601', 'golden_retriever', 0.43086213), ('n02099849', 'Chesapeake_Bay_retriever', 0.2284075), ('n02099712', 'Labrador_retriever', 0.13925594)]\n"
     ]
    }
   ],
   "source": [
    "from keras.applications.resnet50 import ResNet50\n",
    "from keras.preprocessing import image\n",
    "from keras.applications.resnet50 import preprocess_input, decode_predictions\n",
    "import numpy as np\n",
    "\n",
    "model = ResNet50(weights='imagenet')\n",
    "\n",
    "img_path =img_filepath\n",
    "img = image.load_img(img_path, target_size=(224, 224))\n",
    "x = image.img_to_array(img)\n",
    "x = np.expand_dims(x, axis=0)\n",
    "x = preprocess_input(x)\n",
    "\n",
    "preds = model.predict(x)\n",
    "# decode the results into a list of tuples (class, description, probability)\n",
    "# (one such list for each sample in the batch)\n",
    "print('Predicted:', decode_predictions(preds, top=3)[0])"
   ]
  }
 ],
 "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": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}