MNIST Handwritten Digit Classification and Recognition using CNN Deep Learning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf # Import tensorflow library\n",
    "import matplotlib.pyplot as plt # Import matplotlib library\n",
    "import numpy as np # Import numpy library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (60000, 28, 28) y_train shape: (60000,) x_test shape: (10000, 28, 28) y_test shape: (10000,)\n"
     ]
    }
   ],
   "source": [
    "mnist = tf.keras.datasets.mnist # Object of the MNIST dataset\n",
    "(x_train, y_train),(x_test, y_test) = mnist.load_data() # Load data\n",
    "print(\"x_train shape:\", x_train.shape, \"y_train shape:\", y_train.shape,\n",
    "      \"x_test shape:\", x_test.shape, \"y_test shape:\", y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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"
    }
   ],
   "source": [
    "# Show one of the images from the training dataset\n",
    "plt.imshow(x_train[0], cmap=\"gray\") # Import the image\n",
    "plt.xlabel([y_train[0]]) #Add label of the image\n",
    "plt.show() # Plot the image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = tf.keras.utils.normalize(x_train, axis=1) # Normalize the training dataset\n",
    "x_test = tf.keras.utils.normalize(x_test, axis=1) # Normalize the testing dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Build the model object\n",
    "model = tf.keras.models.Sequential()\n",
    "# Add the Flatten Layer\n",
    "model.add(tf.keras.layers.Flatten())\n",
    "# Build the input and the hidden layers\n",
    "model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu))\n",
    "model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu))\n",
    "# Build the output layer\n",
    "model.add(tf.keras.layers.Dense(10, activation=tf.nn.softmax))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\BIPIN\\Anaconda3\\lib\\site-packages\\tensorflow\\python\\ops\\resource_variable_ops.py:435: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "#Compling the model using adam optimization algorithm which is used for training Deep NN.\n",
    "model.compile(optimizer=\"adam\", loss=\"sparse_categorical_crossentropy\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 42000 samples, validate on 18000 samples\n",
      "Epoch 1/5\n",
      "42000/42000 [==============================] - 15s 362us/sample - loss: 0.3593 - acc: 0.8984 - val_loss: 0.1886 - val_acc: 0.9445\n",
      "Epoch 2/5\n",
      "42000/42000 [==============================] - 3s 68us/sample - loss: 0.1429 - acc: 0.9571 - val_loss: 0.1471 - val_acc: 0.9542\n",
      "Epoch 3/5\n",
      "42000/42000 [==============================] - 3s 72us/sample - loss: 0.0973 - acc: 0.9701 - val_loss: 0.1218 - val_acc: 0.9628\n",
      "Epoch 4/5\n",
      " 3600/42000 [=>............................] - ETA: 2s - loss: 0.0709 - acc: 0.9811"
     ]
    }
   ],
   "source": [
    "# Start training process using train data sets\n",
    "model_log=model.fit(x=x_train, y=y_train, batch_size=60, verbose=1, epochs=5, validation_split=.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Evaluate the model performance\n",
    "test_loss, test_acc = model.evaluate(x=x_test, y=y_test)\n",
    "# Print out the model accuracy \n",
    "print('\\nTest Accuracy:', test_acc)\n",
    "print('\\nTest Loss:', test_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = model.predict([x_test]) # Make prediction using test data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(np.argmax(predictions[0])) # Print out the predicted number using index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_test = x_test.reshape(-1,28, 28)\n",
    "x_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Show one of the images from the test dataset\n",
    "plt.imshow(x_test[0],cmap='gray') #Load the image\n",
    "plt.xlabel([y_test[0]]) #Add label of the image\n",
    "plt.show() # Plot the image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "# plotting the graph for accuracy model\n",
    "fig = plt.figure()\n",
    "plt.subplot(2,1,1)\n",
    "plt.plot(model_log.history['acc'])\n",
    "plt.plot(model_log.history['val_acc'])\n",
    "plt.title('Model Accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train Data', 'Test Data'], loc='lower right')\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plotting the graph for loss model\n",
    "plt.subplot(2,1,2)\n",
    "plt.plot(model_log.history['loss'])\n",
    "plt.plot(model_log.history['val_loss'])\n",
    "plt.title('Model Loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train Data', 'Test Data'], loc='upper right')\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_image(i, predictions_array, true_label, img):\n",
    "  predictions_array, true_label, img = predictions_array[i], true_label[i], img[i]\n",
    "  plt.grid(False)\n",
    "  plt.xticks([])\n",
    "  plt.yticks([])\n",
    "  \n",
    "  plt.imshow(img, cmap='gray')\n",
    "\n",
    "  predicted_label = np.argmax(predictions_array)\n",
    "  if predicted_label == true_label:\n",
    "    color = 'blue'\n",
    "  else:\n",
    "    color = 'red'\n",
    "  \n",
    "  plt.xlabel(\"Predicted Label: {} Accuracy: {:1.0f}% True Label: {}\".format([predicted_label],\n",
    "                                100*np.max(predictions_array),\n",
    "                                [true_label]),\n",
    "                                color=color)\n",
    "\n",
    "def plot_value_array(i, predictions_array, true_label):\n",
    "  predictions_array, true_label = predictions_array[i], true_label[i]\n",
    "  plt.grid(False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# assign the number of rows and columns followed with images \n",
    "num_rows = 5\n",
    "num_cols = 3\n",
    "num_images = 9\n",
    "plt.figure(figsize=(2*3*num_cols, 2*num_rows))\n",
    "for i in range(num_images):\n",
    "  plt.subplot(num_rows, 2*num_cols, 2*i+1)\n",
    "  plot_image(i, predictions, y_test, x_test)\n",
    "  \n",
    "plt.show()"
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}