Bird_Classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ao_bChpNLgMV"
   },
   "source": [
    "# **Import Dataset from Kaggle**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "rGdbCEgIkU4e"
   },
   "outputs": [],
   "source": [
    "! pip install --upgrade --force-reinstall --no-deps kaggle\n",
    "from google.colab import files\n",
    "files.upload()\n",
    "! mkdir ~/.kaggle\n",
    "! cp kaggle.json ~/.kaggle/\n",
    "! chmod 600 ~/.kaggle/kaggle.json\n",
    "! kaggle datasets list\n",
    "! kaggle datasets download -d gpiosenka/100-bird-species\n",
    "! unzip 100-bird-species.zip -d dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_-pXBrOTLoG0"
   },
   "source": [
    "# **Import required libraries and modules**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "id": "Kq281wH4kjdF"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow import keras\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras import applications\n",
    "from keras.applications import ResNet152V2\n",
    "from keras.layers import Dropout, BatchNormalization, GlobalAveragePooling2D, Dense\n",
    "from keras.optimizers import Adam"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gyt2JqSSOI8S"
   },
   "source": [
    "# **Paths and Data Augmentation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "id": "1CXtD5H-k3-k"
   },
   "outputs": [],
   "source": [
    "train_path = \"/content/dataset/train\"\n",
    "test_path = \"/content/dataset/test\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "On1_6y5-OR9L"
   },
   "source": [
    "> ImageDataGenerator is used for data augmentation.\n",
    "* Performing horizontal and vertical flip to increase size of dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "id": "Ul2Gyb_JlEhE"
   },
   "outputs": [],
   "source": [
    "datagen = ImageDataGenerator(\n",
    "    rescale = 1./255,\n",
    "    horizontal_flip = True,\n",
    "    vertical_flip = True,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "YSJYK7UilG2D",
    "outputId": "51c6f77f-5174-48b4-9ec0-b29f2c29a6ce"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 35215 images belonging to 250 classes.\n"
     ]
    }
   ],
   "source": [
    "train_data = datagen.flow_from_directory(\n",
    "    train_path,\n",
    "    target_size = (224, 224),\n",
    "    color_mode = 'rgb',\n",
    "    batch_size = 32,\n",
    "    shuffle = True,\n",
    "    seed = 123,\n",
    "    class_mode= 'categorical'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "G62YWQHClJDU",
    "outputId": "d27e2f30-6c01-46c2-c3c2-04a675ad2fcb"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1250 images belonging to 250 classes.\n"
     ]
    }
   ],
   "source": [
    "test_data = datagen.flow_from_directory(\n",
    "    test_path,\n",
    "    target_size = (224,224),\n",
    "    color_mode = 'rgb',\n",
    "    batch_size = 32,\n",
    "    shuffle = True,\n",
    "    seed = 123,\n",
    "    class_mode = 'categorical'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FoHqNUD6PEl5"
   },
   "source": [
    "# **Importing and using Pretrained Model**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jnVATrmqPJ0A"
   },
   "source": [
    "> ResNet152V2\n",
    "\n",
    "* Using pretrained model ResNet 152\n",
    "* Model is pretrained on weights of imagenet\n",
    "* Image input resolution is 224 x 224, colored."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "id": "YT-UCKKilLB7"
   },
   "outputs": [],
   "source": [
    "resnet = ResNet152V2( \n",
    "    include_top = False,\n",
    "    weights = 'imagenet',\n",
    "    input_shape = (224, 224, 3)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "Uqwvg-CJlREU"
   },
   "outputs": [],
   "source": [
    "resnet.trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JxG1jjdMPdZx"
   },
   "source": [
    "# **Creating Model for Classification**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_KEDR2NOPhak"
   },
   "source": [
    "> Sequential Model\n",
    "\n",
    "* Model consists of ResNet for feature extraction.\n",
    "* Dropout layers and BatchNormalization is used to increase accuracy and prevent overfitting\n",
    "* Output layer classifies into 250 distinct classes, as a probability distribution using activation softmax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "id": "-ZSgKW24lpsx"
   },
   "outputs": [],
   "source": [
    "model = tf.keras.Sequential()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "id": "IqG-KjhFlsDU"
   },
   "outputs": [],
   "source": [
    "model.add(resnet)\n",
    "model.add(GlobalAveragePooling2D())\n",
    "model.add(Dropout(0.4))\n",
    "model.add(Dense(2048, activation='relu'))\n",
    "model.add(Dense(256, activation = 'relu'))\n",
    "model.add(Dropout(0.33))\n",
    "model.add(Dense(1024, activation='relu'))\n",
    "model.add(BatchNormalization())\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dense(250, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "VpAk7Z5ImAYF"
   },
   "outputs": [],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vlqjKl9uQGps"
   },
   "source": [
    "# **Compiling and Training the Model**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RmYslvafQJxY"
   },
   "source": [
    "* Model is compiled with optimizer as Adam, with a modified learning rate of 1e-04, as accuracy wasn't great with default learning rate.\n",
    "* Loss function used is categorical crossentropy\n",
    "* Model is trained for 50 epochs, with validation data being the test dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "id": "3xU0th-dmDwX"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer=Adam(learning_rate=0.0001), loss = 'categorical_crossentropy', metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "loRaQtiRmF1D",
    "outputId": "a9ea1272-09de-4f9e-8d78-86f0fae72601"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 1.9296 - accuracy: 0.5029 - val_loss: 0.8227 - val_accuracy: 0.7712\n",
      "Epoch 2/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 1.6188 - accuracy: 0.5696 - val_loss: 0.6519 - val_accuracy: 0.8136\n",
      "Epoch 3/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 1.4350 - accuracy: 0.6115 - val_loss: 0.5648 - val_accuracy: 0.8352\n",
      "Epoch 4/50\n",
      "1101/1101 [==============================] - 269s 245ms/step - loss: 1.3124 - accuracy: 0.6398 - val_loss: 0.5370 - val_accuracy: 0.8304\n",
      "Epoch 5/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 1.2208 - accuracy: 0.6657 - val_loss: 0.4343 - val_accuracy: 0.8720\n",
      "Epoch 6/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 1.1492 - accuracy: 0.6842 - val_loss: 0.4250 - val_accuracy: 0.8816\n",
      "Epoch 7/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 1.0983 - accuracy: 0.6956 - val_loss: 0.3726 - val_accuracy: 0.8816\n",
      "Epoch 8/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 1.0447 - accuracy: 0.7058 - val_loss: 0.3814 - val_accuracy: 0.8816\n",
      "Epoch 9/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 1.0015 - accuracy: 0.7163 - val_loss: 0.3350 - val_accuracy: 0.8920\n",
      "Epoch 10/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.9554 - accuracy: 0.7301 - val_loss: 0.3285 - val_accuracy: 0.9016\n",
      "Epoch 11/50\n",
      "1101/1101 [==============================] - 269s 244ms/step - loss: 0.9177 - accuracy: 0.7401 - val_loss: 0.2920 - val_accuracy: 0.9128\n",
      "Epoch 12/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.8821 - accuracy: 0.7520 - val_loss: 0.2832 - val_accuracy: 0.9184\n",
      "Epoch 13/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.8606 - accuracy: 0.7542 - val_loss: 0.2700 - val_accuracy: 0.9112\n",
      "Epoch 14/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.8289 - accuracy: 0.7619 - val_loss: 0.2836 - val_accuracy: 0.9088\n",
      "Epoch 15/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.7908 - accuracy: 0.7733 - val_loss: 0.2919 - val_accuracy: 0.9080\n",
      "Epoch 16/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.7698 - accuracy: 0.7748 - val_loss: 0.2808 - val_accuracy: 0.9064\n",
      "Epoch 17/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.7548 - accuracy: 0.7828 - val_loss: 0.2542 - val_accuracy: 0.9168\n",
      "Epoch 18/50\n",
      "1101/1101 [==============================] - 271s 247ms/step - loss: 0.7290 - accuracy: 0.7885 - val_loss: 0.2577 - val_accuracy: 0.9208\n",
      "Epoch 19/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.7118 - accuracy: 0.7934 - val_loss: 0.2561 - val_accuracy: 0.9208\n",
      "Epoch 20/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.6905 - accuracy: 0.7968 - val_loss: 0.2555 - val_accuracy: 0.9120\n",
      "Epoch 21/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.6756 - accuracy: 0.8018 - val_loss: 0.2316 - val_accuracy: 0.9312\n",
      "Epoch 22/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.6476 - accuracy: 0.8077 - val_loss: 0.2248 - val_accuracy: 0.9304\n",
      "Epoch 23/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.6307 - accuracy: 0.8153 - val_loss: 0.2411 - val_accuracy: 0.9232\n",
      "Epoch 24/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.6158 - accuracy: 0.8181 - val_loss: 0.2344 - val_accuracy: 0.9264\n",
      "Epoch 25/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.6075 - accuracy: 0.8220 - val_loss: 0.1905 - val_accuracy: 0.9472\n",
      "Epoch 26/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.5879 - accuracy: 0.8253 - val_loss: 0.2428 - val_accuracy: 0.9248\n",
      "Epoch 27/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.5713 - accuracy: 0.8301 - val_loss: 0.2012 - val_accuracy: 0.9368\n",
      "Epoch 28/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.5589 - accuracy: 0.8330 - val_loss: 0.2167 - val_accuracy: 0.9328\n",
      "Epoch 29/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.5480 - accuracy: 0.8346 - val_loss: 0.2228 - val_accuracy: 0.9304\n",
      "Epoch 30/50\n",
      "1101/1101 [==============================] - 271s 247ms/step - loss: 0.5380 - accuracy: 0.8400 - val_loss: 0.1746 - val_accuracy: 0.9448\n",
      "Epoch 31/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.5355 - accuracy: 0.8387 - val_loss: 0.2276 - val_accuracy: 0.9280\n",
      "Epoch 32/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.5152 - accuracy: 0.8466 - val_loss: 0.2026 - val_accuracy: 0.9376\n",
      "Epoch 33/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.5025 - accuracy: 0.8481 - val_loss: 0.1816 - val_accuracy: 0.9480\n",
      "Epoch 34/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.4958 - accuracy: 0.8491 - val_loss: 0.1742 - val_accuracy: 0.9504\n",
      "Epoch 35/50\n",
      "1101/1101 [==============================] - 269s 245ms/step - loss: 0.4810 - accuracy: 0.8535 - val_loss: 0.1741 - val_accuracy: 0.9496\n",
      "Epoch 36/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.4687 - accuracy: 0.8579 - val_loss: 0.1809 - val_accuracy: 0.9384\n",
      "Epoch 37/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.4629 - accuracy: 0.8589 - val_loss: 0.1982 - val_accuracy: 0.9424\n",
      "Epoch 38/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.4600 - accuracy: 0.8605 - val_loss: 0.1989 - val_accuracy: 0.9328\n",
      "Epoch 39/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.4496 - accuracy: 0.8634 - val_loss: 0.2187 - val_accuracy: 0.9256\n",
      "Epoch 40/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.4341 - accuracy: 0.8652 - val_loss: 0.2086 - val_accuracy: 0.9336\n",
      "Epoch 41/50\n",
      "1101/1101 [==============================] - 269s 245ms/step - loss: 0.4323 - accuracy: 0.8692 - val_loss: 0.1897 - val_accuracy: 0.9384\n",
      "Epoch 42/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.4268 - accuracy: 0.8684 - val_loss: 0.1840 - val_accuracy: 0.9360\n",
      "Epoch 43/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.4088 - accuracy: 0.8742 - val_loss: 0.1723 - val_accuracy: 0.9440\n",
      "Epoch 44/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.4155 - accuracy: 0.8731 - val_loss: 0.1843 - val_accuracy: 0.9400\n",
      "Epoch 45/50\n",
      "1101/1101 [==============================] - 270s 245ms/step - loss: 0.3940 - accuracy: 0.8760 - val_loss: 0.1914 - val_accuracy: 0.9384\n",
      "Epoch 46/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.3928 - accuracy: 0.8784 - val_loss: 0.1885 - val_accuracy: 0.9456\n",
      "Epoch 47/50\n",
      "1101/1101 [==============================] - 272s 247ms/step - loss: 0.3829 - accuracy: 0.8817 - val_loss: 0.1852 - val_accuracy: 0.9464\n",
      "Epoch 48/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.3888 - accuracy: 0.8801 - val_loss: 0.1784 - val_accuracy: 0.9448\n",
      "Epoch 49/50\n",
      "1101/1101 [==============================] - 270s 246ms/step - loss: 0.3716 - accuracy: 0.8851 - val_loss: 0.1856 - val_accuracy: 0.9488\n",
      "Epoch 50/50\n",
      "1101/1101 [==============================] - 271s 246ms/step - loss: 0.3664 - accuracy: 0.8876 - val_loss: 0.2068 - val_accuracy: 0.9448\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(train_data, validation_data=test_data, epochs = 50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RU9GNUbeQh_6"
   },
   "source": [
    "# **Plotting Graphs**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "A9t6948DQmoO"
   },
   "source": [
    "> **Using Matplotlib Library**\n",
    "* Graph is plotted for ``` Epoch vs Accuracy ```, for training accuracy and validation accuracy.\n",
    "* Graph is plotted for ``` Epoch vs Losses ```, for training loss and validation loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 296
    },
    "id": "jiYKz6MIokAZ",
    "outputId": "1e20f09f-bd5c-439f-af07-5258801d8d50"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f15d0268a10>"
      ]
     },
     "execution_count": 23,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['accuracy'], label = 'Accuracy')\n",
    "plt.plot(history.history['val_accuracy'], label = 'Validation Accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend(loc = \"lower right\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 298
    },
    "id": "ZkvyGYlNol9W",
    "outputId": "7751339b-c56d-4b07-a0fd-5b8b96edc335"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f1523ed3050>"
      ]
     },
     "execution_count": 50,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'], label = 'Loss')\n",
    "plt.plot(history.history['val_loss'], label = 'Validation Loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc = \"upper right\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "V7wCpZA3RH7N"
   },
   "source": [
    "# **Saving the model as HDF5 File**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "id": "ORBqQA3SwsLZ"
   },
   "outputs": [],
   "source": [
    "model.save('birds.h5')"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Bird-Classification.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "tf",
   "language": "python",
   "name": "tf"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}