Kaggle CNN Cancer Benchmark 95% accuracy.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from keras_preprocessing.image import ImageDataGenerator\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Activation, Flatten, Dropout,BatchNormalization\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "from keras import regularizers, optimizers\n",
    "from keras.metrics import categorical_accuracy\n",
    "from keras.callbacks import EarlyStopping, ReduceLROnPlateau\n",
    "import time"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Label Read In & Prep"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(1)\n",
    "\n",
    "df = pd.read_csv(r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection/train_labels.csv')\n",
    "\n",
    "def append_ext(fn):\n",
    "    return fn+\".tif\"\n",
    "\n",
    "df[\"id\"]=df[\"id\"].apply(append_ext)\n",
    "                   \n",
    "\n",
    "df[\"label\"]=df[\"label\"].astype(str)\n",
    "\n",
    "train_path = r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection\\train'\n",
    "valid_path = r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection\\train'\n",
    "\n",
    "testdf= pd.read_csv(r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection/sample_submission.csv')\n",
    "\n",
    "def append_ext2(fn):\n",
    "    return r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection\\test/' + fn+\".tif\"\n",
    "\n",
    "testdf[\"id\"]=testdf[\"id\"].apply(append_ext2)\n",
    "                   \n",
    "                   \n",
    "testdf = testdf.drop(['label'],axis=1)\n",
    "\n",
    "test_path = r'C:\\Users\\casocha\\Downloads\\histopathologic-cancer-detection\\test/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Data Read In"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 57458 images.\n",
      "Found 187022 images belonging to 2 classes.\n",
      "Found 33003 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "test_generator=test_datagen.flow_from_dataframe(\n",
    "dataframe=testdf,\n",
    "directory=\"test_path\",\n",
    "x_col=\"id\",\n",
    "y_col=None,\n",
    "batch_size=128,\n",
    "seed=1,\n",
    "shuffle=False,\n",
    "class_mode=None,\n",
    "use_mutiprocessing = True,\n",
    "target_size=(96,96))\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "       # horizontal_flip=True,\n",
    "       #vertical_flip=True,\n",
    "       #brightness_range=[0.5, 1.5],\n",
    "       #fill_mode='reflect',                               \n",
    "        #rotation_range=15,\n",
    "        rescale=1./255,\n",
    "        #shear_range=0.2,\n",
    "        #zoom_range=0.2\n",
    "        validation_split=0.15\n",
    "    \n",
    ")\n",
    "\n",
    "\n",
    "train_generator = train_datagen.flow_from_dataframe(\n",
    "                dataframe=df,\n",
    "                directory=train_path,\n",
    "                x_col = 'id',\n",
    "                y_col = 'label',\n",
    "                has_ext=False,\n",
    "                subset='training',\n",
    "                target_size=(96, 96),\n",
    "                batch_size=128,\n",
    "                seed=1,\n",
    "                class_mode='binary',\n",
    "                use_mutiprocessing = True\n",
    "                )\n",
    "\n",
    "validation_generator = train_datagen.flow_from_dataframe(\n",
    "                dataframe=df,\n",
    "                directory=valid_path,\n",
    "                x_col = 'id',\n",
    "                y_col = 'label',\n",
    "                has_ext=False,\n",
    "                subset='validation', # This is the trick to properly separate train and validation dataset\n",
    "                target_size=(96, 96),\n",
    "                batch_size=128,\n",
    "                seed=1,\n",
    "                shuffle=False,\n",
    "                class_mode='binary',\n",
    "                use_mutiprocessing = True\n",
    "                )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Baseline Model Definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_112 (Conv2D)          (None, 96, 96, 32)        896       \n",
      "_________________________________________________________________\n",
      "activation_127 (Activation)  (None, 96, 96, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_113 (Conv2D)          (None, 96, 96, 32)        9248      \n",
      "_________________________________________________________________\n",
      "activation_128 (Activation)  (None, 96, 96, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_114 (Conv2D)          (None, 96, 96, 32)        9248      \n",
      "_________________________________________________________________\n",
      "activation_129 (Activation)  (None, 96, 96, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_37 (MaxPooling (None, 48, 48, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_115 (Conv2D)          (None, 48, 48, 64)        18496     \n",
      "_________________________________________________________________\n",
      "activation_130 (Activation)  (None, 48, 48, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_116 (Conv2D)          (None, 48, 48, 64)        36928     \n",
      "_________________________________________________________________\n",
      "activation_131 (Activation)  (None, 48, 48, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_117 (Conv2D)          (None, 48, 48, 64)        36928     \n",
      "_________________________________________________________________\n",
      "activation_132 (Activation)  (None, 48, 48, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_38 (MaxPooling (None, 24, 24, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_118 (Conv2D)          (None, 24, 24, 128)       73856     \n",
      "_________________________________________________________________\n",
      "activation_133 (Activation)  (None, 24, 24, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_119 (Conv2D)          (None, 24, 24, 128)       147584    \n",
      "_________________________________________________________________\n",
      "activation_134 (Activation)  (None, 24, 24, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_120 (Conv2D)          (None, 24, 24, 128)       147584    \n",
      "_________________________________________________________________\n",
      "activation_135 (Activation)  (None, 24, 24, 128)       0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_39 (MaxPooling (None, 12, 12, 128)       0         \n",
      "_________________________________________________________________\n",
      "flatten_14 (Flatten)         (None, 18432)             0         \n",
      "_________________________________________________________________\n",
      "dense_30 (Dense)             (None, 256)               4718848   \n",
      "_________________________________________________________________\n",
      "activation_136 (Activation)  (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_31 (Dense)             (None, 512)               131584    \n",
      "_________________________________________________________________\n",
      "activation_137 (Activation)  (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_32 (Dense)             (None, 1)                 513       \n",
      "=================================================================\n",
      "Total params: 5,331,713\n",
      "Trainable params: 5,331,713\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "##Inspired by the VGG architecture \n",
    "model = Sequential()\n",
    "model.add(Conv2D(32,(3,3),padding='same',input_shape=(96,96,3)))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(32, (3, 3),padding='same'))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(32, (3, 3),padding='same'))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "\n",
    "model.add(Conv2D(64,(3,3),padding='same'))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(64, (3, 3),padding='same'))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(64, (3, 3),padding='same'))\n",
    "#model.add(BatchNormalization())\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "\n",
    "model.add(Conv2D(128,(3,3),padding='same'))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(128, (3, 3),padding='same'))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Conv2D(128, (3, 3),padding='same'))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "\n",
    "model.add(Flatten())\n",
    "model.add(Dense(256))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dense(512))\n",
    "model.add(Activation('relu'))\n",
    "#model.add(Dropout(0.3))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.summary()\n",
    "model.compile(optimizer='adam',loss=\"binary_crossentropy\",metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/6\n",
      "1241/1461 [========================>.....] - ETA: 2:36 - loss: 0.4527 - acc: 0.7811"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda\\lib\\site-packages\\keras\\callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.133355). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1461/1461 [==============================] - 1094s 749ms/step - loss: 0.4389 - acc: 0.7903 - val_loss: 0.3671 - val_acc: 0.8373\n",
      "Epoch 2/6\n",
      "1461/1461 [==============================] - 1090s 746ms/step - loss: 0.3058 - acc: 0.8713 - val_loss: 0.2591 - val_acc: 0.8961\n",
      "Epoch 3/6\n",
      "1243/1461 [========================>.....] - ETA: 2:33 - loss: 0.2504 - acc: 0.8984"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda\\lib\\site-packages\\keras\\callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.156268). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1461/1461 [==============================] - 1088s 745ms/step - loss: 0.2446 - acc: 0.9011 - val_loss: 0.2125 - val_acc: 0.9180\n",
      "Epoch 4/6\n",
      "1244/1461 [========================>.....] - ETA: 2:33 - loss: 0.2130 - acc: 0.9155"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda\\lib\\site-packages\\keras\\callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.109374). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1461/1461 [==============================] - 1090s 746ms/step - loss: 0.2108 - acc: 0.9167 - val_loss: 0.2003 - val_acc: 0.9234\n",
      "Epoch 5/6\n",
      "1245/1461 [========================>.....] - ETA: 2:32 - loss: 0.1896 - acc: 0.9274"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda\\lib\\site-packages\\keras\\callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.148967). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1461/1461 [==============================] - 1091s 747ms/step - loss: 0.1870 - acc: 0.9285 - val_loss: 0.2026 - val_acc: 0.9237\n",
      "\n",
      "Epoch 00005: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.\n",
      "Epoch 6/6\n",
      "1246/1461 [========================>.....] - ETA: 2:31 - loss: 0.1364 - acc: 0.9502"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda\\lib\\site-packages\\keras\\callbacks.py:122: UserWarning: Method on_batch_end() is slow compared to the batch update (0.188042). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1461/1461 [==============================] - 1090s 746ms/step - loss: 0.1336 - acc: 0.9513 - val_loss: 0.1622 - val_acc: 0.9424\n",
      "Training time:  109.0954150835673  minutes\n"
     ]
    }
   ],
   "source": [
    "STEP_SIZE_TRAIN=train_generator.n//train_generator.batch_size\n",
    "STEP_SIZE_VALID=validation_generator.n//validation_generator.batch_size\n",
    "STEP_SIZE_TEST=test_generator.n//test_generator.batch_size\n",
    "\n",
    "earlystopper = EarlyStopping(monitor='val_loss', patience=2, verbose=1, restore_best_weights=True)\n",
    "reducel = ReduceLROnPlateau(monitor='val_loss', patience=1, verbose=1, factor=0.1)\n",
    "\n",
    "start = time.time()\n",
    "history = model.fit_generator(generator=train_generator,\n",
    "                    steps_per_epoch=STEP_SIZE_TRAIN,\n",
    "                    validation_data=validation_generator,\n",
    "                    validation_steps=STEP_SIZE_VALID,\n",
    "                    epochs=6,callbacks=[reducel, earlystopper]\n",
    ")\n",
    "\n",
    "end = time.time()\n",
    "print(\"Training time: \", (end - start)/60, \" minutes\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.16948769339637876, 0.9393960539919445]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate_generator(generator=validation_generator,\n",
    "steps=STEP_SIZE_TEST)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.figure(figsize=(12,6))\n",
    "# summarize history for accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.title('Adam w/o BatchNormalization Accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Adam w/o BatchNormalization Loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "fp = r'Downloads\\models/cancer-base.hdf5'\n",
    "model.save(fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": 2
}