2-efficientnetB0_Custom_dataset.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EfficientNet Implementation   - EfficientNet- B0\n",
    "\n",
    "## Training a Custom  Model from scratch"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Pre Processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['earphone', 'headphone', 'phones']\n",
      "Types of classes labels found:  3\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "import numpy as np \n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "\n",
    "dataset_path = os.listdir('dataset')\n",
    "\n",
    "print (dataset_path)  #what kinds of classes are in this dataset\n",
    "\n",
    "print(\"Types of classes labels found: \", len(dataset_path))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_labels = []\n",
    "\n",
    "for item in dataset_path:\n",
    " # Get all the file names\n",
    " all_classes = os.listdir('dataset' + '/' +item)\n",
    " #print(all_classes)\n",
    "\n",
    " # Add them to the list\n",
    " for room in all_classes:\n",
    "    class_labels.append((item, str('dataset_path' + '/' +item) + '/' + room))\n",
    "    #print(class_labels[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Labels                         image\n",
      "0  earphone  dataset_path/earphone/21.png\n",
      "1  earphone  dataset_path/earphone/22.png\n",
      "2  earphone  dataset_path/earphone/23.png\n",
      "3  earphone  dataset_path/earphone/24.png\n",
      "4  earphone  dataset_path/earphone/25.png\n",
      "     Labels                           image\n",
      "98   phones  dataset_path/phones/phs112.png\n",
      "99   phones  dataset_path/phones/phs123.jpg\n",
      "100  phones  dataset_path/phones/phs123.png\n",
      "101  phones  dataset_path/phones/phs157.png\n",
      "102  phones  dataset_path/phones/phs165.png\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Build a dataframe        \n",
    "df = pd.DataFrame(data=class_labels, columns=['Labels', 'image'])\n",
    "print(df.head())\n",
    "print(df.tail())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of images in the dataset:  103\n",
      "headphone    50\n",
      "phones       29\n",
      "earphone     24\n",
      "Name: Labels, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Let's check how many samples for each category are present\n",
    "print(\"Total number of images in the dataset: \", len(df))\n",
    "\n",
    "label_count = df['Labels'].value_counts()\n",
    "print(label_count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "path = 'dataset/'\n",
    "dataset_path = os.listdir('dataset')\n",
    "\n",
    "im_size = 224\n",
    "\n",
    "images = []\n",
    "labels = []\n",
    "\n",
    "for i in dataset_path:\n",
    "    data_path = path + str(i)  \n",
    "    filenames = [i for i in os.listdir(data_path) ]\n",
    "   \n",
    "    for f in filenames:\n",
    "        img = cv2.imread(data_path + '/' + f)\n",
    "        img = cv2.resize(img, (im_size, im_size))\n",
    "        images.append(img)\n",
    "        labels.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(103, 224, 224, 3)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "#This model takes input images of shape (224, 224, 3), and the input data should range [0, 255]. \n",
    "\n",
    "images = np.array(images)\n",
    "\n",
    "images = images.astype('float32') / 255.0\n",
    "images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['earphone' 'earphone' 'earphone' 'earphone' 'earphone' 'earphone'\n",
      " 'earphone' 'earphone' 'earphone' 'earphone' 'earphone' 'earphone'\n",
      " 'earphone' 'earphone' 'earphone' 'earphone' 'earphone' 'earphone'\n",
      " 'earphone' 'earphone' 'earphone' 'earphone' 'earphone' 'earphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'headphone' 'headphone' 'headphone' 'headphone'\n",
      " 'headphone' 'headphone' 'phones' 'phones' 'phones' 'phones' 'phones'\n",
      " 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones'\n",
      " 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones'\n",
      " 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones' 'phones']\n",
      "[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      " 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      " 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder , OneHotEncoder\n",
    "y=df['Labels'].values\n",
    "print(y)\n",
    "\n",
    "y_labelencoder = LabelEncoder ()\n",
    "y = y_labelencoder.fit_transform (y)\n",
    "print (y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]]\n",
      "[[0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "y=y.reshape(-1,1)\n",
    "\n",
    "from sklearn.compose import ColumnTransformer\n",
    "ct = ColumnTransformer([('my_ohe', OneHotEncoder(), [0])], remainder='passthrough')\n",
    "Y = ct.fit_transform(y) #.toarray()\n",
    "print(Y[:5])\n",
    "print(Y[35:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(97, 224, 224, 3)\n",
      "(97, 3)\n",
      "(6, 224, 224, 3)\n",
      "(6, 3)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.utils import shuffle\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "images, Y = shuffle(images, Y, random_state=1)\n",
    "\n",
    "\n",
    "train_x, test_x, train_y, test_y = train_test_split(images, Y, test_size=0.05, random_state=415)\n",
    "\n",
    "#inpect the shape of the training and testing.\n",
    "print(train_x.shape)\n",
    "print(train_y.shape)\n",
    "print(test_x.shape)\n",
    "print(test_y.shape)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "# EfficientNet Implementation :\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.applications import EfficientNetB0\n",
    "\n",
    "NUM_CLASSES = 3\n",
    "IMG_SIZE = 224\n",
    "size = (IMG_SIZE, IMG_SIZE)\n",
    "\n",
    "\n",
    "inputs = layers.Input(shape=(IMG_SIZE, IMG_SIZE, 3))\n",
    "\n",
    "\n",
    "# Using model without transfer learning\n",
    "\n",
    "outputs = EfficientNetB0(include_top=True, weights=None, classes=NUM_CLASSES)(inputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 224, 224, 3)]     0         \n",
      "_________________________________________________________________\n",
      "efficientnetb0 (Functional)  (None, 3)                 4053414   \n",
      "=================================================================\n",
      "Total params: 4,053,414\n",
      "Trainable params: 4,011,391\n",
      "Non-trainable params: 42,023\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.4590s vs `on_train_batch_end` time: 0.8895s). Check your callbacks.\n",
      "4/4 - 6s - loss: 8.2172 - accuracy: 0.3402\n",
      "Epoch 2/30\n",
      "4/4 - 5s - loss: 3.5772 - accuracy: 0.2887\n",
      "Epoch 3/30\n",
      "4/4 - 4s - loss: 3.6683 - accuracy: 0.4124\n",
      "Epoch 4/30\n",
      "4/4 - 3s - loss: 2.9720 - accuracy: 0.5773\n",
      "Epoch 5/30\n",
      "4/4 - 1s - loss: 1.6091 - accuracy: 0.6598\n",
      "Epoch 6/30\n",
      "4/4 - 1s - loss: 1.6499 - accuracy: 0.6186\n",
      "Epoch 7/30\n",
      "4/4 - 1s - loss: 0.8478 - accuracy: 0.6598\n",
      "Epoch 8/30\n",
      "4/4 - 1s - loss: 0.8477 - accuracy: 0.7216\n",
      "Epoch 9/30\n",
      "4/4 - 1s - loss: 0.6787 - accuracy: 0.8247\n",
      "Epoch 10/30\n",
      "4/4 - 1s - loss: 0.9582 - accuracy: 0.7629\n",
      "Epoch 11/30\n",
      "4/4 - 1s - loss: 0.3642 - accuracy: 0.8763\n",
      "Epoch 12/30\n",
      "4/4 - 1s - loss: 0.6927 - accuracy: 0.8247\n",
      "Epoch 13/30\n",
      "4/4 - 1s - loss: 0.2888 - accuracy: 0.8866\n",
      "Epoch 14/30\n",
      "4/4 - 1s - loss: 0.3690 - accuracy: 0.8763\n",
      "Epoch 15/30\n",
      "4/4 - 1s - loss: 0.4281 - accuracy: 0.8763\n",
      "Epoch 16/30\n",
      "4/4 - 1s - loss: 0.3383 - accuracy: 0.8866\n",
      "Epoch 17/30\n",
      "4/4 - 1s - loss: 0.3477 - accuracy: 0.9072\n",
      "Epoch 18/30\n",
      "4/4 - 1s - loss: 0.5455 - accuracy: 0.8866\n",
      "Epoch 19/30\n",
      "4/4 - 1s - loss: 0.3235 - accuracy: 0.8969\n",
      "Epoch 20/30\n",
      "4/4 - 1s - loss: 0.0948 - accuracy: 0.9588\n",
      "Epoch 21/30\n",
      "4/4 - 1s - loss: 0.1610 - accuracy: 0.9381\n",
      "Epoch 22/30\n",
      "4/4 - 1s - loss: 0.1860 - accuracy: 0.9278\n",
      "Epoch 23/30\n",
      "4/4 - 1s - loss: 0.5188 - accuracy: 0.9278\n",
      "Epoch 24/30\n",
      "4/4 - 1s - loss: 0.1059 - accuracy: 0.9691\n",
      "Epoch 25/30\n",
      "4/4 - 1s - loss: 0.6584 - accuracy: 0.8144\n",
      "Epoch 26/30\n",
      "4/4 - 1s - loss: 0.3449 - accuracy: 0.8969\n",
      "Epoch 27/30\n",
      "4/4 - 1s - loss: 0.4066 - accuracy: 0.8866\n",
      "Epoch 28/30\n",
      "4/4 - 1s - loss: 0.5110 - accuracy: 0.8557\n",
      "Epoch 29/30\n",
      "4/4 - 1s - loss: 0.2519 - accuracy: 0.9278\n",
      "Epoch 30/30\n",
      "4/4 - 1s - loss: 0.2225 - accuracy: 0.9175\n"
     ]
    }
   ],
   "source": [
    "model = tf.keras.Model(inputs, outputs)\n",
    "\n",
    "model.compile(optimizer=\"adam\", loss=\"categorical_crossentropy\", metrics=[\"accuracy\"] )\n",
    "\n",
    "model.summary()\n",
    "\n",
    "hist = model.fit(train_x, train_y, epochs=30, verbose=2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "def plot_hist(hist):\n",
    "    plt.plot(hist.history[\"accuracy\"])\n",
    "    #plt.plot(hist.history[\"val_accuracy\"])\n",
    "    plt.title(\"model accuracy\")\n",
    "    plt.ylabel(\"accuracy\")\n",
    "    plt.xlabel(\"epoch\")\n",
    "    plt.legend([\"train\", \"validation\"], loc=\"upper left\")\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "plot_hist(hist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 2ms/step - loss: 1.7834 - accuracy: 0.3333\n",
      "Loss = 1.7833582162857056\n",
      "Test Accuracy = 0.3333333432674408\n"
     ]
    }
   ],
   "source": [
    "preds = model.evaluate(test_x, test_y)\n",
    "print (\"Loss = \" + str(preds[0]))\n",
    "print (\"Test Accuracy = \" + str(preds[1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing Efficient Model On Unseen data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input image shape: (1, 224, 224, 3)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1b1f61a2448>"
      ]
     },
     "execution_count": 14,
     "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": [
    "from matplotlib.pyplot import imread\n",
    "from matplotlib.pyplot import imshow\n",
    "from tensorflow.keras.preprocessing import image\n",
    "from tensorflow.keras.applications.imagenet_utils import decode_predictions\n",
    "from tensorflow.keras.applications.imagenet_utils import preprocess_input\n",
    "\n",
    "\n",
    "img_path = 'unseen.jfif'\n",
    "\n",
    "#img = image.load_img(img_path, target_size=(224, 224))\n",
    "#x = img.img_to_array(img)\n",
    "\n",
    "img = cv2.imread(img_path)\n",
    "img = cv2.resize(img, (224, 224))\n",
    "\n",
    "x = np.expand_dims(img, axis=0)\n",
    "x = preprocess_input(x)\n",
    "\n",
    "print('Input image shape:', x.shape)\n",
    "\n",
    "my_image = imread(img_path)\n",
    "imshow(my_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.01885055, 0.2305208 , 0.7506286 ]], dtype=float32)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds=model.predict(x)\n",
    "preds     # probabilities for being in each of the 3 classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PhysicalDevice(name='/physical_device:CPU:0', device_type='CPU'),\n",
       " PhysicalDevice(name='/physical_device:XLA_CPU:0', device_type='XLA_CPU'),\n",
       " PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'),\n",
       " PhysicalDevice(name='/physical_device:XLA_GPU:0', device_type='XLA_GPU')]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Cuda and cudnn is installed for this tensorflow version. So we can see GPU is enabled\n",
    "tf.config.experimental.list_physical_devices()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "4/4 - 10s - loss: 0.0314 - accuracy: 0.9794\n",
      "Epoch 2/30\n",
      "4/4 - 10s - loss: 0.0112 - accuracy: 1.0000\n",
      "Epoch 3/30\n",
      "4/4 - 10s - loss: 0.0575 - accuracy: 0.9794\n",
      "Epoch 4/30\n",
      "4/4 - 10s - loss: 0.0127 - accuracy: 1.0000\n",
      "Epoch 5/30\n",
      "4/4 - 10s - loss: 0.1307 - accuracy: 0.9691\n",
      "Epoch 6/30\n",
      "4/4 - 9s - loss: 0.1140 - accuracy: 0.9588\n",
      "Epoch 7/30\n",
      "4/4 - 10s - loss: 0.1884 - accuracy: 0.9485\n",
      "Epoch 8/30\n",
      "4/4 - 10s - loss: 0.1041 - accuracy: 0.9485\n",
      "Epoch 9/30\n",
      "4/4 - 9s - loss: 0.1310 - accuracy: 0.9794\n",
      "Epoch 10/30\n",
      "4/4 - 10s - loss: 0.1590 - accuracy: 0.9485\n",
      "Epoch 11/30\n",
      "4/4 - 10s - loss: 0.0269 - accuracy: 0.9897\n",
      "Epoch 12/30\n",
      "4/4 - 10s - loss: 0.3513 - accuracy: 0.9381\n",
      "Epoch 13/30\n",
      "4/4 - 10s - loss: 0.0182 - accuracy: 0.9897\n",
      "Epoch 14/30\n",
      "4/4 - 10s - loss: 0.1573 - accuracy: 0.9278\n",
      "Epoch 15/30\n",
      "4/4 - 10s - loss: 0.1249 - accuracy: 0.9588\n",
      "Epoch 16/30\n",
      "4/4 - 10s - loss: 0.2126 - accuracy: 0.9278\n",
      "Epoch 17/30\n",
      "4/4 - 10s - loss: 0.0770 - accuracy: 0.9691\n",
      "Epoch 18/30\n",
      "4/4 - 10s - loss: 0.0753 - accuracy: 0.9691\n",
      "Epoch 19/30\n",
      "4/4 - 10s - loss: 0.0125 - accuracy: 1.0000\n",
      "Epoch 20/30\n",
      "4/4 - 10s - loss: 0.2589 - accuracy: 0.9691\n",
      "Epoch 21/30\n",
      "4/4 - 10s - loss: 0.2965 - accuracy: 0.9072\n",
      "Epoch 22/30\n",
      "4/4 - 10s - loss: 0.2434 - accuracy: 0.9278\n",
      "Epoch 23/30\n",
      "4/4 - 11s - loss: 0.1788 - accuracy: 0.9381\n",
      "Epoch 24/30\n",
      "4/4 - 10s - loss: 0.4289 - accuracy: 0.9175\n",
      "Epoch 25/30\n",
      "4/4 - 10s - loss: 0.5336 - accuracy: 0.8969\n",
      "Epoch 26/30\n",
      "4/4 - 10s - loss: 0.2831 - accuracy: 0.8660\n",
      "Epoch 27/30\n",
      "4/4 - 16s - loss: 0.1846 - accuracy: 0.9588\n",
      "Epoch 28/30\n",
      "4/4 - 10s - loss: 0.1903 - accuracy: 0.9485\n",
      "Epoch 29/30\n",
      "4/4 - 14s - loss: 0.1100 - accuracy: 0.9588\n",
      "Epoch 30/30\n",
      "4/4 - 15s - loss: 0.1196 - accuracy: 0.9691\n",
      "7min 53s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit -n1 -r1 \n",
    "with tf.device('/CPU:0'):\n",
    "    cpu_performance =model.fit(train_x, train_y, epochs=30, verbose=2)\n",
    "    cpu_performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.0710s vs `on_train_batch_end` time: 0.1970s). Check your callbacks.\n",
      "4/4 - 1s - loss: 0.2818 - accuracy: 0.9278\n",
      "Epoch 2/30\n",
      "4/4 - 1s - loss: 0.1793 - accuracy: 0.9588\n",
      "Epoch 3/30\n",
      "4/4 - 1s - loss: 0.1113 - accuracy: 0.9588\n",
      "Epoch 4/30\n",
      "4/4 - 1s - loss: 0.1415 - accuracy: 0.9588\n",
      "Epoch 5/30\n",
      "4/4 - 1s - loss: 0.3826 - accuracy: 0.9485\n",
      "Epoch 6/30\n",
      "4/4 - 1s - loss: 0.0562 - accuracy: 0.9794\n",
      "Epoch 7/30\n",
      "4/4 - 1s - loss: 0.1397 - accuracy: 0.9691\n",
      "Epoch 8/30\n",
      "4/4 - 1s - loss: 0.2192 - accuracy: 0.9072\n",
      "Epoch 9/30\n",
      "4/4 - 1s - loss: 0.0741 - accuracy: 0.9588\n",
      "Epoch 10/30\n",
      "4/4 - 1s - loss: 0.3673 - accuracy: 0.9072\n",
      "Epoch 11/30\n",
      "4/4 - 1s - loss: 0.0513 - accuracy: 0.9794\n",
      "Epoch 12/30\n",
      "4/4 - 1s - loss: 0.0676 - accuracy: 0.9897\n",
      "Epoch 13/30\n",
      "4/4 - 1s - loss: 0.0743 - accuracy: 0.9794\n",
      "Epoch 14/30\n",
      "4/4 - 1s - loss: 0.0084 - accuracy: 1.0000\n",
      "Epoch 15/30\n",
      "4/4 - 1s - loss: 0.0847 - accuracy: 0.9691\n",
      "Epoch 16/30\n",
      "4/4 - 1s - loss: 0.0291 - accuracy: 0.9897\n",
      "Epoch 17/30\n",
      "4/4 - 1s - loss: 0.1700 - accuracy: 0.9381\n",
      "Epoch 18/30\n",
      "4/4 - 1s - loss: 0.0905 - accuracy: 0.9691\n",
      "Epoch 19/30\n",
      "4/4 - 1s - loss: 0.0990 - accuracy: 0.9485\n",
      "Epoch 20/30\n",
      "4/4 - 1s - loss: 0.0211 - accuracy: 1.0000\n",
      "Epoch 21/30\n",
      "4/4 - 1s - loss: 0.0792 - accuracy: 0.9691\n",
      "Epoch 22/30\n",
      "4/4 - 1s - loss: 0.0382 - accuracy: 0.9794\n",
      "Epoch 23/30\n",
      "4/4 - 1s - loss: 0.0165 - accuracy: 0.9897\n",
      "Epoch 24/30\n",
      "4/4 - 1s - loss: 0.0193 - accuracy: 0.9897\n",
      "Epoch 25/30\n",
      "4/4 - 1s - loss: 0.0490 - accuracy: 0.9794\n",
      "Epoch 26/30\n",
      "4/4 - 1s - loss: 0.0581 - accuracy: 0.9794\n",
      "Epoch 27/30\n",
      "4/4 - 1s - loss: 0.0185 - accuracy: 0.9897\n",
      "Epoch 28/30\n",
      "4/4 - 1s - loss: 0.0509 - accuracy: 0.9794\n",
      "Epoch 29/30\n",
      "4/4 - 1s - loss: 0.0035 - accuracy: 1.0000\n",
      "Epoch 30/30\n",
      "4/4 - 1s - loss: 0.0310 - accuracy: 0.9897\n",
      "25.6 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "%%timeit -n1 -r1 \n",
    "with tf.device('/GPU:0'):\n",
    "    gpu_performance =model.fit(train_x, train_y, epochs=30, verbose=2)\n",
    "    gpu_performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# CPU completed the training in 7 min 53 Seconds and GPU did that training in 25.6 seconds"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}