model1-Conv_2-batchnormalisation-augmentation-3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.callbacks import *\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.models import Sequential\n",
    "from keras.layers import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata = pd.read_csv('sign_mnist_train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainlabel=traindata['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata.drop('label',inplace=True,axis=1)\n",
    "trainimages = traindata.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainimages=trainimages.reshape(-1,28,28,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "testdata = pd.read_csv('sign_mnist_test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "testlabel=testdata['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "testdata.drop('label',inplace=True,axis=1)\n",
    "testimages = testdata.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "testimages=testimages.reshape(-1,28,28,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "traingen=ImageDataGenerator(rotation_range=40,zoom_range=0.1,width_shift_range=0.1,height_shift_range=0.1,\n",
    "                  shear_range=0.1,horizontal_flip=True,rescale=1/255.0,validation_split=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "traindata_generator = traingen.flow(trainimages,trainlabel,subset='training')\n",
    "validationdata_generator = traingen.flow(trainimages,trainlabel,subset='validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "testgen=ImageDataGenerator(rescale=1/255.0)\n",
    "testdata_generator = testgen.flow(testimages,testlabel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "model=Sequential([])\n",
    "\n",
    "model.add(Conv2D(64,(3,3),activation=\"relu\",input_shape=(28,28,1)))\n",
    "model.add(BatchNormalization())\n",
    "# model.add(Conv2D(64,(3,3),activation=\"relu\"))\n",
    "# model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(2,2))\n",
    "#model.add(Dropout(0.4))\n",
    "\n",
    "model.add(Conv2D(128,(3,3),activation=\"relu\"))\n",
    "model.add(BatchNormalization())\n",
    "# model.add(Conv2D(128,(3,3),activation=\"relu\"))\n",
    "# model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(2,2))\n",
    "#model.add(Dropout(0.4))\n",
    "\n",
    "model.add(Flatten())\n",
    "model.add(BatchNormalization())\n",
    "#model.add(Dropout(0.4))\n",
    "model.add(Dense(256,activation=\"relu\"))\n",
    "model.add(BatchNormalization())\n",
    "#model.add(Dropout(0.4))\n",
    "model.add(Dense(26,activation=\"softmax\"))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\",optimizer='adam',metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_3 (Conv2D)            (None, 26, 26, 64)        640       \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 26, 26, 64)        256       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 13, 13, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 11, 11, 128)       73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 11, 11, 128)       512       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 5, 5, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 3200)              0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_7 (Batch (None, 3200)              12800     \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 256)               819456    \n",
      "_________________________________________________________________\n",
      "batch_normalization_8 (Batch (None, 256)               1024      \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 26)                6682      \n",
      "=================================================================\n",
      "Total params: 915,226\n",
      "Trainable params: 907,930\n",
      "Non-trainable params: 7,296\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define a Callback class that stops training once accuracy reaches 97.0%\n",
    "class myCallback(Callback):\n",
    "  def on_epoch_end(self, epoch, logs={}):\n",
    "    if(logs.get('val_accuracy')>0.995):\n",
    "      print(\"\\nReached 99.5% accuracy so cancelling training!\")\n",
    "      self.model.stop_training = True\n",
    "callback=myCallback()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "773/773 [==============================] - 9s 11ms/step - loss: 1.2847 - accuracy: 0.5999 - val_loss: 1.8255 - val_accuracy: 0.6098\n",
      "Epoch 2/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.4966 - accuracy: 0.8390 - val_loss: 0.7818 - val_accuracy: 0.7166\n",
      "Epoch 3/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.3288 - accuracy: 0.8934 - val_loss: 0.6771 - val_accuracy: 0.8149\n",
      "Epoch 4/60\n",
      "773/773 [==============================] - 9s 11ms/step - loss: 0.2607 - accuracy: 0.9140 - val_loss: 0.5003 - val_accuracy: 0.8273\n",
      "Epoch 5/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.2205 - accuracy: 0.9287 - val_loss: 0.6071 - val_accuracy: 0.8485\n",
      "Epoch 6/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1870 - accuracy: 0.9409 - val_loss: 0.4184 - val_accuracy: 0.8204\n",
      "Epoch 7/60\n",
      "773/773 [==============================] - 9s 11ms/step - loss: 0.1661 - accuracy: 0.9466 - val_loss: 3.3362 - val_accuracy: 0.7144\n",
      "Epoch 8/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1437 - accuracy: 0.9512 - val_loss: 0.7521 - val_accuracy: 0.8135\n",
      "Epoch 9/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1661 - accuracy: 0.9442 - val_loss: 1.9637 - val_accuracy: 0.7399\n",
      "Epoch 10/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1331 - accuracy: 0.9558 - val_loss: 0.3046 - val_accuracy: 0.9199\n",
      "Epoch 11/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1123 - accuracy: 0.9632 - val_loss: 154.5653 - val_accuracy: 0.0648\n",
      "Epoch 12/60\n",
      "773/773 [==============================] - 9s 11ms/step - loss: 0.1243 - accuracy: 0.9596 - val_loss: 0.0240 - val_accuracy: 0.9621\n",
      "Epoch 13/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1063 - accuracy: 0.9653 - val_loss: 0.1127 - val_accuracy: 0.9199\n",
      "Epoch 14/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1006 - accuracy: 0.9658 - val_loss: 1.7908 - val_accuracy: 0.9614\n",
      "Epoch 15/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0998 - accuracy: 0.9660 - val_loss: 0.1100 - val_accuracy: 0.9741\n",
      "Epoch 16/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0895 - accuracy: 0.9699 - val_loss: 0.5922 - val_accuracy: 0.8634\n",
      "Epoch 17/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.1025 - accuracy: 0.9673 - val_loss: 0.1131 - val_accuracy: 0.8638\n",
      "Epoch 18/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0901 - accuracy: 0.9696 - val_loss: 0.0699 - val_accuracy: 0.9563\n",
      "Epoch 19/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0832 - accuracy: 0.9725 - val_loss: 71.3070 - val_accuracy: 0.1607\n",
      "Epoch 20/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0840 - accuracy: 0.9725 - val_loss: 0.1330 - val_accuracy: 0.9741\n",
      "Epoch 21/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0797 - accuracy: 0.9736 - val_loss: 77.2878 - val_accuracy: 0.1705\n",
      "Epoch 22/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0853 - accuracy: 0.9718 - val_loss: 0.2152 - val_accuracy: 0.9770\n",
      "Epoch 23/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0728 - accuracy: 0.9757 - val_loss: 0.0098 - val_accuracy: 0.9683\n",
      "Epoch 24/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0671 - accuracy: 0.9781 - val_loss: 11.6072 - val_accuracy: 0.7228\n",
      "Epoch 25/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0757 - accuracy: 0.9756 - val_loss: 0.1040 - val_accuracy: 0.9413\n",
      "Epoch 26/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0638 - accuracy: 0.9785 - val_loss: 1.2606 - val_accuracy: 0.7202\n",
      "Epoch 27/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0731 - accuracy: 0.9766 - val_loss: 0.0562 - val_accuracy: 0.9807\n",
      "Epoch 28/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0671 - accuracy: 0.9771 - val_loss: 0.5320 - val_accuracy: 0.9344\n",
      "Epoch 29/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0581 - accuracy: 0.9805 - val_loss: 2.1538 - val_accuracy: 0.8182\n",
      "Epoch 30/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0623 - accuracy: 0.9787 - val_loss: 0.1125 - val_accuracy: 0.9158\n",
      "Epoch 31/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0706 - accuracy: 0.9773 - val_loss: 0.0902 - val_accuracy: 0.9730\n",
      "Epoch 32/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0568 - accuracy: 0.9817 - val_loss: 0.0012 - val_accuracy: 0.9865\n",
      "Epoch 33/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0634 - accuracy: 0.9810 - val_loss: 0.0278 - val_accuracy: 0.9650\n",
      "Epoch 34/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0517 - accuracy: 0.9840 - val_loss: 14.4931 - val_accuracy: 0.4838\n",
      "Epoch 35/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0567 - accuracy: 0.9828 - val_loss: 157.0176 - val_accuracy: 0.0721\n",
      "Epoch 36/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0731 - accuracy: 0.9769 - val_loss: 0.0183 - val_accuracy: 0.9880\n",
      "Epoch 37/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0427 - accuracy: 0.9848 - val_loss: 0.0088 - val_accuracy: 0.9770\n",
      "Epoch 38/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0493 - accuracy: 0.9837 - val_loss: 0.1569 - val_accuracy: 0.9184\n",
      "Epoch 39/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0555 - accuracy: 0.9815 - val_loss: 0.0364 - val_accuracy: 0.9515\n",
      "Epoch 40/60\n",
      "773/773 [==============================] - 9s 11ms/step - loss: 0.0552 - accuracy: 0.9830 - val_loss: 0.1180 - val_accuracy: 0.9297\n",
      "Epoch 41/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0399 - accuracy: 0.9864 - val_loss: 4.4171 - val_accuracy: 0.6871\n",
      "Epoch 42/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0528 - accuracy: 0.9828 - val_loss: 0.0018 - val_accuracy: 0.9709\n",
      "Epoch 43/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0606 - accuracy: 0.9805 - val_loss: 0.0041 - val_accuracy: 0.9359\n",
      "Epoch 44/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0497 - accuracy: 0.9829 - val_loss: 0.4110 - val_accuracy: 0.9304\n",
      "Epoch 45/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0456 - accuracy: 0.9846 - val_loss: 0.0505 - val_accuracy: 0.9741\n",
      "Epoch 46/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0416 - accuracy: 0.9856 - val_loss: 0.0424 - val_accuracy: 0.9625\n",
      "Epoch 47/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0393 - accuracy: 0.9872 - val_loss: 0.0259 - val_accuracy: 0.9508\n",
      "Epoch 48/60\n",
      "773/773 [==============================] - 9s 12ms/step - loss: 0.0406 - accuracy: 0.9864 - val_loss: 0.0137 - val_accuracy: 0.9862\n",
      "Epoch 49/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0468 - accuracy: 0.9845 - val_loss: 0.0441 - val_accuracy: 0.9836\n",
      "Epoch 50/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0364 - accuracy: 0.9873 - val_loss: 0.1408 - val_accuracy: 0.9279\n",
      "Epoch 51/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0397 - accuracy: 0.9868 - val_loss: 0.4303 - val_accuracy: 0.8153\n",
      "Epoch 52/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0399 - accuracy: 0.9858 - val_loss: 910.1866 - val_accuracy: 0.0189\n",
      "Epoch 53/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0412 - accuracy: 0.9861 - val_loss: 0.1814 - val_accuracy: 0.9749\n",
      "Epoch 54/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0363 - accuracy: 0.9884 - val_loss: 0.0468 - val_accuracy: 0.9650\n",
      "Epoch 55/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0347 - accuracy: 0.9888 - val_loss: 6.1029 - val_accuracy: 0.8230\n",
      "Epoch 56/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0394 - accuracy: 0.9876 - val_loss: 0.0032 - val_accuracy: 0.9905\n",
      "Epoch 57/60\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "773/773 [==============================] - 9s 12ms/step - loss: 0.0379 - accuracy: 0.9884 - val_loss: 0.0048 - val_accuracy: 0.9938\n",
      "Epoch 58/60\n",
      "773/773 [==============================] - 8s 11ms/step - loss: 0.0429 - accuracy: 0.9860 - val_loss: 0.5717 - val_accuracy: 0.8834\n",
      "Epoch 59/60\n",
      "773/773 [==============================] - 8s 10ms/step - loss: 0.0351 - accuracy: 0.9882 - val_loss: 0.7017 - val_accuracy: 0.8597\n",
      "Epoch 60/60\n",
      "773/773 [==============================] - 9s 12ms/step - loss: 0.0396 - accuracy: 0.9870 - val_loss: 0.0083 - val_accuracy: 0.9851\n"
     ]
    }
   ],
   "source": [
    "history=model.fit(traindata_generator,epochs=60,validation_data=validationdata_generator,callbacks=[callback])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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": [
    "acc = history.history['accuracy']\n",
    "val_acc = history.history['val_accuracy']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'r', label='Training accuracy')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'r', label='Training Loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation Loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test accuracy: 99.51199293136597\n"
     ]
    }
   ],
   "source": [
    "print(\"test accuracy: \"+ str(model.evaluate_generator(testdata_generator)[1]*100))"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}