Object Detection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyMnYMvSmP/Moyt+sZtyuL9O",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/patel-deepak/obejectdetection/blob/main/Object%20Detection.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**2138921_DEEPAK_PATEL_CSCM45J - Coursework 2: Object Recognition**"
      ],
      "metadata": {
        "id": "L8Nr77YTsfpw"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> LOAD ALL Dataset files in colab: \n",
        "\n",
        "1. trnLabel_coarse.npy\n",
        "2. trnLabel_fine.npy\n",
        "3. trnImage.npy  \n",
        "4. tstImage.npy\n",
        "5. tstLabel_coarse.npy\n",
        "6. tstLabel_fine.npy \n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "ie5qNwVbs2Tu"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 73
        },
        "id": "v7MfS-a35f59",
        "outputId": "6c356d34-0dc1-47e0-86be-de0eb96e85cf"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-520b3a80-7265-4ab9-830d-76a16151785a\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-520b3a80-7265-4ab9-830d-76a16151785a\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving tstImage.npy to tstImage.npy\n"
          ]
        }
      ],
      "source": [
        "from google.colab import files\n",
        "uploaded = files.upload()"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import skimage.feature"
      ],
      "metadata": {
        "id": "Vv9B14TOQWVL"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Load datasets with numpy and check the shape of the data"
      ],
      "metadata": {
        "id": "0AigQs5Mt9Km"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "images = np.load('./trnImage.npy')\n",
        "label_fine = np.load('./trnLabel_fine.npy')\n",
        "label_coarse = np.load('./trnLabel_coarse.npy')\n",
        "\n",
        "print(f'Images Shape: {images.shape}')\n",
        "print(f'Images Fine Labels Shape: {label_fine.shape}')\n",
        "print(f'Images Coarse Labels Shape: {label_coarse.shape}')\n",
        "image_index = 2 # pick a specific image\n",
        "image = images[:, :, :, image_index]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Pm68aWqLQaw_",
        "outputId": "3350076e-affe-4347-b01c-20b00eef0207"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Images Shape: (32, 32, 3, 50000)\n",
            "Images Fine Labels Shape: (50000,)\n",
            "Images Coarse Labels Shape: (50000,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> As mentioned in RunME verify reshaped ,Transpose and Looped data in order to use it for Machine Learning .\n",
        "As per this verification I will be using transpose data for further processing."
      ],
      "metadata": {
        "id": "JcjXyEjWuHVY"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Lets call reshape on our 4D tensor\n",
        "reshaped = np.reshape(images, (50000, 32, 32, 3))\n",
        "print(f'Shape of reshaped: {reshaped.shape}')\n",
        "\n",
        "# Lets call transpose on our 4D tensor\n",
        "transposed = np.transpose(images, [3,0,1,2])\n",
        "print(f'Shape of transposed: {transposed.shape}')\n",
        "\n",
        "# Lets try looping over our 4D tensor, note how long this takes!\n",
        "new_images = []\n",
        "for i in range(images.shape[-1]):\n",
        "    new_images.append(images[:,:,:,i])\n",
        "new_images = np.asarray(new_images)\n",
        "print(f'Shape of new_images: {new_images.shape}')\n",
        "\n",
        "# Now lets look at what the result is\n",
        "plt.figure()\n",
        "plt.imshow(images[:,:,:,0])\n",
        "plt.title('Original Images')\n",
        "plt.show()\n",
        "plt.figure()\n",
        "plt.imshow(reshaped[0])\n",
        "plt.title('Reshaped')\n",
        "plt.show()\n",
        "plt.figure()\n",
        "plt.imshow(transposed[0])\n",
        "plt.title('Transposed')\n",
        "plt.show()\n",
        "plt.figure()\n",
        "plt.imshow(new_images[0])\n",
        "plt.title('Looped')\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "jIOLxlfvQgqD",
        "outputId": "d21d614a-8cb8-4aaf-ea76-6aae70894920"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Shape of reshaped: (50000, 32, 32, 3)\n",
            "Shape of transposed: (50000, 32, 32, 3)\n",
            "Shape of new_images: (50000, 32, 32, 3)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> As per RunMe extract the feature and view it ."
      ],
      "metadata": {
        "id": "gF6TX8XHuize"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Extract features from a single image\n",
        "hog_feature, hog_image = skimage.feature.hog(image, pixels_per_cell=[6,6], cells_per_block=[2,2], visualize=True) # you may need to add \"channel_axis=-1\" if you are using the newest version of skimage\n",
        "\n",
        "print(f'The extracted feature vector is of length {hog_feature.shape[0]} per sample')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HeIufirLQ4On",
        "outputId": "0cc9c0a9-3330-4198-c06a-3a151b3d4d02"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "The extracted feature vector is of length 576 per sample\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure()\n",
        "plt.subplot(121)\n",
        "plt.imshow(hog_image)\n",
        "plt.subplot(122)\n",
        "plt.imshow(image)\n",
        "plt.suptitle(f'Extracted HOG features from image number: {image_index}\\nFine Class: {label_fine[image_index]}, Coarse Class: {label_coarse[image_index]}')\n",
        "plt.show(block=False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 262
        },
        "id": "IC6dVX2DQ7W4",
        "outputId": "aec25950-cb7e-4960-dd05-6d12a1068812"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf"
      ],
      "metadata": {
        "id": "UD8jQg3oRUJD"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Just to check how much time it takes to load the data directly from cifar100 dataset.\n",
        "this process takes very less time as compared to loading I did in start of the process using colab upload process."
      ],
      "metadata": {
        "id": "c49lqiz3usEJ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "trn, tst = tf.keras.datasets.cifar100.load_data()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8PkRh8TiRXbu",
        "outputId": "5422ee29-56c3-4772-c13f-a91414a903a8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-100-python.tar.gz\n",
            "169009152/169001437 [==============================] - 11s 0us/step\n",
            "169017344/169001437 [==============================] - 11s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "for j in range(65,100):\n",
        "    res = np.where(trn[1] == j)[0]\n",
        "\n",
        "    for i in range(2):\n",
        "        plt.figure()\n",
        "        plt.imshow(trn[0][res[i]])"
      ],
      "metadata": {
        "id": "WfJV28_BRay1"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> This section is to check what is tstLabel_fine and tstLabel_coarse\n",
        "\n",
        "\n",
        "1.   tstLabel_coarse - its a superclass of 20 \n",
        "2.   tstLabel_fine   - its a class of 100\n",
        "\n"
      ],
      "metadata": {
        "id": "MtcsAoPsvIAK"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "images_T = np.load('./tstImage.npy')\n",
        "label_fine_T = np.load('./tstLabel_fine.npy')\n",
        "label_coarse_T = np.load('./tstLabel_coarse.npy')\n",
        "\n",
        "\n",
        "# Lets call transpose on our 4D tensor\n",
        "transposed_T = np.transpose(images_T, [3,0,1,2])\n",
        "\n",
        "\n",
        "print(transposed.shape)\n",
        "print(label_fine.shape)\n",
        "print(label_coarse.shape)\n",
        "\n",
        "print(transposed_T.shape)\n",
        "print(label_fine_T.shape)\n",
        "print(label_coarse_T.shape)\n",
        "\n",
        "print(max(label_fine_T))   # it's a  class\n",
        "print(max(label_fine))     # it's a  class\n",
        "print(max(label_coarse_T)) # it's a super class\n",
        "print(max(label_coarse))   # it's a super class\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d-ydrVXWSAII",
        "outputId": "4c098b78-6c63-48ee-dfd2-b5f6ab7a0b2d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(50000, 32, 32, 3)\n",
            "(50000,)\n",
            "(50000,)\n",
            "(10000, 32, 32, 3)\n",
            "(10000,)\n",
            "(10000,)\n",
            "99\n",
            "99\n",
            "19\n",
            "19\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(transposed[0])\n",
        "print(label_fine[0])\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "o-vdqLmkUk8M",
        "outputId": "e7ea6d10-e506-440d-fc8d-9ea15bbd0e2f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "19\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(transposed_T[0])\n",
        "print(label_fine_T[0])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "yXLgDtpOVDDu",
        "outputId": "3b8ecedf-f6e0-48aa-fdbf-18e5f0455f2f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "49\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> first tried to train the model with normalised data by dividing 255 but the resutls was not good so kept the data as is ."
      ],
      "metadata": {
        "id": "2v5Azg21wmg0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "data = transposed /1\n",
        "data_T = transposed_T /1"
      ],
      "metadata": {
        "id": "_GoLT8_IThyE"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.python.ops.gen_nn_ops import Conv2D\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf"
      ],
      "metadata": {
        "id": "iDRnfxMlTuur"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Built ANN with dense layers of 1024 and 100 neurons for label of classes(100).training accuracy was very less it's only 22%.So didn't  proceed further."
      ],
      "metadata": {
        "id": "kjpbnWmv4MQN"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model = tf.keras.Sequential(layers=[\n",
        "        tf.keras.layers.Flatten(input_shape=(32,32,3)), \n",
        "        tf.keras.layers.Dense(1024,activation='relu',),\n",
        "        tf.keras.layers.Dense(100,activation=tf.nn.softmax)])\n",
        "\n",
        "model.compile(optimizer=tf.keras.optimizers.Adam(), \n",
        "              loss=tf.keras.losses.SparseCategoricalCrossentropy(), \n",
        "              metrics=tf.keras.metrics.SparseCategoricalAccuracy())\n",
        "\n",
        "history = model.fit(data,label_fine, epochs=15) "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HFmG5XazaWxj",
        "outputId": "5fbd5ac5-00f5-464a-e96c-23e7e5d83846"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/15\n",
            "1563/1563 [==============================] - 38s 24ms/step - loss: 4.2294 - sparse_categorical_accuracy: 0.0620\n",
            "Epoch 2/15\n",
            "1563/1563 [==============================] - 39s 25ms/step - loss: 3.8991 - sparse_categorical_accuracy: 0.1098\n",
            "Epoch 3/15\n",
            "1563/1563 [==============================] - 39s 25ms/step - loss: 3.7566 - sparse_categorical_accuracy: 0.1326\n",
            "Epoch 4/15\n",
            "1563/1563 [==============================] - 38s 24ms/step - loss: 3.6696 - sparse_categorical_accuracy: 0.1480\n",
            "Epoch 5/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.6050 - sparse_categorical_accuracy: 0.1597\n",
            "Epoch 6/15\n",
            "1563/1563 [==============================] - 38s 24ms/step - loss: 3.5551 - sparse_categorical_accuracy: 0.1697\n",
            "Epoch 7/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.5109 - sparse_categorical_accuracy: 0.1764\n",
            "Epoch 8/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.4722 - sparse_categorical_accuracy: 0.1858\n",
            "Epoch 9/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.4384 - sparse_categorical_accuracy: 0.1913\n",
            "Epoch 10/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.4063 - sparse_categorical_accuracy: 0.1985\n",
            "Epoch 11/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.3765 - sparse_categorical_accuracy: 0.2039\n",
            "Epoch 12/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.3495 - sparse_categorical_accuracy: 0.2083\n",
            "Epoch 13/15\n",
            "1563/1563 [==============================] - 37s 24ms/step - loss: 3.3234 - sparse_categorical_accuracy: 0.2131\n",
            "Epoch 14/15\n",
            "1563/1563 [==============================] - 38s 25ms/step - loss: 3.2976 - sparse_categorical_accuracy: 0.2188\n",
            "Epoch 15/15\n",
            "1563/1563 [==============================] - 40s 26ms/step - loss: 3.2726 - sparse_categorical_accuracy: 0.2216\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Built CNN with Conv2D layer, poolMaxPooling2D  layer ,and dense layers of 3000 and 100 neurons for label of classes(100).\n",
        "\n",
        "Using this model training accuracy is : 92% for 100 classes.\n",
        "Using this model testing accuracy is : 24.83% for 100 classes."
      ],
      "metadata": {
        "id": "7lhZQeE24QZE"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model = tf.keras.models.Sequential([\n",
        "    tf.keras.layers.Conv2D(64, (7,7), activation='relu', input_shape=(32, 32, 3)),\n",
        "    tf.keras.layers.MaxPooling2D(2, 2),\n",
        "    tf.keras.layers.Conv2D(32, (5, 5), activation='relu'),\n",
        "    tf.keras.layers.MaxPooling2D(2, 2),\n",
        "    tf.keras.layers.Flatten(),\n",
        "    tf.keras.layers.Dense(3000, activation='relu'),\n",
        "    tf.keras.layers.Dense(100, activation='softmax')\n",
        "])\n",
        "\n",
        "model.summary()\n",
        "\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "\n",
        "history = model.fit(data,label_fine, epochs=25)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YS2AnNV1fHwH",
        "outputId": "23289221-5214-45b1-d943-56b7bfea0db4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_13\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d_26 (Conv2D)          (None, 26, 26, 64)        9472      \n",
            "                                                                 \n",
            " max_pooling2d_26 (MaxPoolin  (None, 13, 13, 64)       0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " conv2d_27 (Conv2D)          (None, 9, 9, 32)          51232     \n",
            "                                                                 \n",
            " max_pooling2d_27 (MaxPoolin  (None, 4, 4, 32)         0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " flatten_14 (Flatten)        (None, 512)               0         \n",
            "                                                                 \n",
            " dense_26 (Dense)            (None, 3000)              1539000   \n",
            "                                                                 \n",
            " dense_27 (Dense)            (None, 100)               300100    \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 1,899,804\n",
            "Trainable params: 1,899,804\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 3.8814 - accuracy: 0.1078\n",
            "Epoch 2/25\n",
            "1563/1563 [==============================] - 153s 98ms/step - loss: 3.2911 - accuracy: 0.2083\n",
            "Epoch 3/25\n",
            "1563/1563 [==============================] - 154s 98ms/step - loss: 2.9926 - accuracy: 0.2637\n",
            "Epoch 4/25\n",
            "1563/1563 [==============================] - 151s 97ms/step - loss: 2.7425 - accuracy: 0.3114\n",
            "Epoch 5/25\n",
            "1563/1563 [==============================] - 152s 98ms/step - loss: 2.4922 - accuracy: 0.3629\n",
            "Epoch 6/25\n",
            "1563/1563 [==============================] - 151s 97ms/step - loss: 2.2262 - accuracy: 0.4158\n",
            "Epoch 7/25\n",
            "1563/1563 [==============================] - 151s 96ms/step - loss: 1.9412 - accuracy: 0.4823\n",
            "Epoch 8/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 1.6568 - accuracy: 0.5496\n",
            "Epoch 9/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 1.3918 - accuracy: 0.6133\n",
            "Epoch 10/25\n",
            "1563/1563 [==============================] - 151s 96ms/step - loss: 1.1540 - accuracy: 0.6725\n",
            "Epoch 11/25\n",
            "1563/1563 [==============================] - 151s 97ms/step - loss: 0.9549 - accuracy: 0.7238\n",
            "Epoch 12/25\n",
            "1563/1563 [==============================] - 150s 96ms/step - loss: 0.7951 - accuracy: 0.7677\n",
            "Epoch 13/25\n",
            "1563/1563 [==============================] - 150s 96ms/step - loss: 0.6827 - accuracy: 0.7989\n",
            "Epoch 14/25\n",
            "1563/1563 [==============================] - 154s 99ms/step - loss: 0.5927 - accuracy: 0.8229\n",
            "Epoch 15/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 0.5147 - accuracy: 0.8476\n",
            "Epoch 16/25\n",
            "1563/1563 [==============================] - 151s 97ms/step - loss: 0.4541 - accuracy: 0.8634\n",
            "Epoch 17/25\n",
            "1563/1563 [==============================] - 153s 98ms/step - loss: 0.4452 - accuracy: 0.8683\n",
            "Epoch 18/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 0.3861 - accuracy: 0.8876\n",
            "Epoch 19/25\n",
            "1563/1563 [==============================] - 152s 98ms/step - loss: 0.3633 - accuracy: 0.8921\n",
            "Epoch 20/25\n",
            "1563/1563 [==============================] - 153s 98ms/step - loss: 0.3611 - accuracy: 0.8934\n",
            "Epoch 21/25\n",
            "1563/1563 [==============================] - 154s 98ms/step - loss: 0.3272 - accuracy: 0.9049\n",
            "Epoch 22/25\n",
            "1563/1563 [==============================] - 153s 98ms/step - loss: 0.3075 - accuracy: 0.9124\n",
            "Epoch 23/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 0.3120 - accuracy: 0.9119\n",
            "Epoch 24/25\n",
            "1563/1563 [==============================] - 153s 98ms/step - loss: 0.2966 - accuracy: 0.9157\n",
            "Epoch 25/25\n",
            "1563/1563 [==============================] - 152s 97ms/step - loss: 0.2712 - accuracy: 0.9229\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.evaluate(data_T,label_fine_T)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dGdxgNsRT0iC",
        "outputId": "6cd3621c-9163-42d7-e698-0338808423e3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 8s 26ms/step - loss: 9.7875 - accuracy: 0.2483\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[9.787501335144043, 0.2483000010251999]"
            ]
          },
          "metadata": {},
          "execution_count": 43
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Built CNN with Conv2D layer, poolMaxPooling2D  layer ,and dense layers of 1024 and 100 neurons for label of super classes(20).\n",
        "\n",
        "Using this model training accuracy is : 71.77% for 20 super classes.\n",
        "Using this model testing accuracy is  : 36.66% for 20 super classes."
      ],
      "metadata": {
        "id": "YuF2cl5A4t6X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model1 = tf.keras.models.Sequential([\n",
        "    tf.keras.layers.Conv2D(64, (7,7), activation='relu', input_shape=(32, 32, 3)),\n",
        "    tf.keras.layers.MaxPooling2D(2, 2),\n",
        "    tf.keras.layers.Conv2D(32, (5, 5), activation='relu'),\n",
        "    tf.keras.layers.MaxPooling2D(2, 2),\n",
        "    tf.keras.layers.Flatten(),\n",
        "    tf.keras.layers.Dense(512, activation='relu'),\n",
        "    tf.keras.layers.Dense(20, activation='softmax')\n",
        "])\n",
        "\n",
        "model1.summary()\n",
        "\n",
        "model1.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "\n",
        "history = model1.fit(data,label_coarse, epochs=20)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8alSEBRkx_bi",
        "outputId": "200640ab-7c90-4d6e-8782-ba69eb5ab07a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d (Conv2D)             (None, 26, 26, 64)        9472      \n",
            "                                                                 \n",
            " max_pooling2d (MaxPooling2D  (None, 13, 13, 64)       0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv2d_1 (Conv2D)           (None, 9, 9, 32)          51232     \n",
            "                                                                 \n",
            " max_pooling2d_1 (MaxPooling  (None, 4, 4, 32)         0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 512)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 512)               262656    \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 20)                10260     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 333,620\n",
            "Trainable params: 333,620\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/20\n",
            "1563/1563 [==============================] - 128s 81ms/step - loss: 2.5507 - accuracy: 0.2090\n",
            "Epoch 2/20\n",
            "1563/1563 [==============================] - 131s 84ms/step - loss: 2.2446 - accuracy: 0.3063\n",
            "Epoch 3/20\n",
            "1563/1563 [==============================] - 128s 82ms/step - loss: 2.1065 - accuracy: 0.3487\n",
            "Epoch 4/20\n",
            "1563/1563 [==============================] - 129s 82ms/step - loss: 2.0055 - accuracy: 0.3774\n",
            "Epoch 5/20\n",
            "1563/1563 [==============================] - 130s 83ms/step - loss: 1.9086 - accuracy: 0.4069\n",
            "Epoch 6/20\n",
            "1563/1563 [==============================] - 130s 83ms/step - loss: 1.8195 - accuracy: 0.4331\n",
            "Epoch 7/20\n",
            "1563/1563 [==============================] - 135s 86ms/step - loss: 1.7302 - accuracy: 0.4596\n",
            "Epoch 8/20\n",
            "1563/1563 [==============================] - 130s 83ms/step - loss: 1.6486 - accuracy: 0.4850\n",
            "Epoch 9/20\n",
            "1563/1563 [==============================] - 130s 83ms/step - loss: 1.5658 - accuracy: 0.5084\n",
            "Epoch 10/20\n",
            "1563/1563 [==============================] - 130s 83ms/step - loss: 1.4892 - accuracy: 0.5309\n",
            "Epoch 11/20\n",
            "1563/1563 [==============================] - 131s 84ms/step - loss: 1.4141 - accuracy: 0.5530\n",
            "Epoch 12/20\n",
            "1563/1563 [==============================] - 131s 84ms/step - loss: 1.3371 - accuracy: 0.5761\n",
            "Epoch 13/20\n",
            "1563/1563 [==============================] - 131s 84ms/step - loss: 1.2668 - accuracy: 0.6003\n",
            "Epoch 14/20\n",
            "1563/1563 [==============================] - 131s 84ms/step - loss: 1.1982 - accuracy: 0.6190\n",
            "Epoch 15/20\n",
            "1563/1563 [==============================] - 132s 85ms/step - loss: 1.1315 - accuracy: 0.6402\n",
            "Epoch 16/20\n",
            "1563/1563 [==============================] - 132s 85ms/step - loss: 1.0670 - accuracy: 0.6582\n",
            "Epoch 17/20\n",
            "1563/1563 [==============================] - 133s 85ms/step - loss: 1.0135 - accuracy: 0.6775\n",
            "Epoch 18/20\n",
            "1563/1563 [==============================] - 134s 86ms/step - loss: 0.9611 - accuracy: 0.6918\n",
            "Epoch 19/20\n",
            "1563/1563 [==============================] - 133s 85ms/step - loss: 0.9127 - accuracy: 0.7084\n",
            "Epoch 20/20\n",
            "1563/1563 [==============================] - 137s 87ms/step - loss: 0.8744 - accuracy: 0.7177\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model1.evaluate(data_T,label_coarse_T)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sQbSoqzqU_lX",
        "outputId": "e0ba6544-93fe-4f2a-9cb5-9815c8109dcf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 8s 21ms/step - loss: 3.0508 - accuracy: 0.3667\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[3.0507636070251465, 0.3666999936103821]"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Created another Model for both type of lebels using hog extracted images.**"
      ],
      "metadata": {
        "id": "-Xg21-saV0mk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "hog_feature, hog_image = skimage.feature.hog(image, pixels_per_cell=[6,6], cells_per_block=[2,2], visualize=True) # you may need to add \"channel_axis=-1\" if you are using the newest version of skimage\n",
        "\n",
        "print(f'The extracted feature vector is of length {hog_feature.shape[0]} per sample')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZBI_wX28Xqc-",
        "outputId": "9de46adb-37fb-4db0-b029-5e949ddcf112"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "The extracted feature vector is of length 576 per sample\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(hog_feature.shape)\n",
        "print(hog_image.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Gcbb1fNqWhd5",
        "outputId": "1a0c405a-a05a-41fd-ed68-a0729c7c5f99"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(576,)\n",
            "(32, 32)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==> Extract the features using hog function."
      ],
      "metadata": {
        "id": "9q-hgg-X5Evm"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "n_training = len( data )\n",
        "n_testing = len( data_T )\n",
        "normalize = True          #  True ==> yields a little bit better score\n",
        "                          #  \n",
        "block_norm = 'L2-Hys'     # or 'L1'\n",
        "orientations = 9          # \n",
        "pixels_per_cell = [8, 8]  #  see section 'Additional remarks' for some explanation\n",
        "cells_per_block = [2, 2]  # \n",
        "\n",
        "def extractFeature(img, vis=False):\n",
        "    from skimage.feature import hog\n",
        "    return hog(img, orientations, pixels_per_cell, cells_per_block, block_norm, visualize=vis, transform_sqrt=normalize)\n",
        "# feature extraction\n",
        "import numpy as np\n",
        "X_training = np.array( [ extractFeature(data[i], vis=False) for i in range(n_training) ] )\n",
        "X_testing = np.array( [ extractFeature(data_T[i], vis=False) for i in range(n_testing) ] )\n"
      ],
      "metadata": {
        "id": "eVkLNssYYV0w"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(X_training.shape)\n",
        "print(X_testing.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5l575YfjY2Jc",
        "outputId": "972b7547-ae87-4abc-ae4a-27325db4c3b3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(50000, 324)\n",
            "(10000, 324)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==>  Built ANN model using extracted features for 100 classses. \n",
        "For this model training accuracy is 49.35 %\n",
        "For this model testing accuracy is 23.56 %"
      ],
      "metadata": {
        "id": "3dSnhWuA5PKO"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2 = tf.keras.Sequential(layers=[\n",
        "        tf.keras.layers.Dense(324,activation='relu',),\n",
        "#        tf.keras.layers.Dense(200,activation='relu',),\n",
        "        tf.keras.layers.Dense(100,activation=tf.nn.softmax)])\n",
        "\n",
        "model2.compile(optimizer=tf.keras.optimizers.Adam(), \n",
        "              loss=tf.keras.losses.SparseCategoricalCrossentropy(), \n",
        "              metrics=tf.keras.metrics.SparseCategoricalAccuracy())\n",
        "\n",
        "history = model2.fit(X_training,label_fine, epochs=50) "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ncYEX5ZeZ8XK",
        "outputId": "8cd7394c-56c9-42e7-9b35-fc958ee72d5d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 3.8717 - sparse_categorical_accuracy: 0.1287\n",
            "Epoch 2/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 3.4209 - sparse_categorical_accuracy: 0.2023\n",
            "Epoch 3/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 3.2693 - sparse_categorical_accuracy: 0.2288\n",
            "Epoch 4/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 3.1645 - sparse_categorical_accuracy: 0.2492\n",
            "Epoch 5/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 3.0801 - sparse_categorical_accuracy: 0.2649\n",
            "Epoch 6/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 3.0060 - sparse_categorical_accuracy: 0.2779\n",
            "Epoch 7/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.9406 - sparse_categorical_accuracy: 0.2906\n",
            "Epoch 8/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.8800 - sparse_categorical_accuracy: 0.3008\n",
            "Epoch 9/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.8280 - sparse_categorical_accuracy: 0.3123\n",
            "Epoch 10/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.7779 - sparse_categorical_accuracy: 0.3222\n",
            "Epoch 11/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.7324 - sparse_categorical_accuracy: 0.3302\n",
            "Epoch 12/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.6898 - sparse_categorical_accuracy: 0.3393\n",
            "Epoch 13/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.6475 - sparse_categorical_accuracy: 0.3478\n",
            "Epoch 14/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.6106 - sparse_categorical_accuracy: 0.3539\n",
            "Epoch 15/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.5757 - sparse_categorical_accuracy: 0.3599\n",
            "Epoch 16/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.5399 - sparse_categorical_accuracy: 0.3685\n",
            "Epoch 17/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.5074 - sparse_categorical_accuracy: 0.3749\n",
            "Epoch 18/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.4763 - sparse_categorical_accuracy: 0.3801\n",
            "Epoch 19/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.4496 - sparse_categorical_accuracy: 0.3868\n",
            "Epoch 20/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.4229 - sparse_categorical_accuracy: 0.3913\n",
            "Epoch 21/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3969 - sparse_categorical_accuracy: 0.3951\n",
            "Epoch 22/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3723 - sparse_categorical_accuracy: 0.4036\n",
            "Epoch 23/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3459 - sparse_categorical_accuracy: 0.4071\n",
            "Epoch 24/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3244 - sparse_categorical_accuracy: 0.4108\n",
            "Epoch 25/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3019 - sparse_categorical_accuracy: 0.4150\n",
            "Epoch 26/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.2809 - sparse_categorical_accuracy: 0.4194\n",
            "Epoch 27/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.2614 - sparse_categorical_accuracy: 0.4257\n",
            "Epoch 28/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.2416 - sparse_categorical_accuracy: 0.4304\n",
            "Epoch 29/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.2217 - sparse_categorical_accuracy: 0.4344\n",
            "Epoch 30/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.2049 - sparse_categorical_accuracy: 0.4357\n",
            "Epoch 31/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1883 - sparse_categorical_accuracy: 0.4389\n",
            "Epoch 32/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1704 - sparse_categorical_accuracy: 0.4446\n",
            "Epoch 33/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1550 - sparse_categorical_accuracy: 0.4476\n",
            "Epoch 34/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1370 - sparse_categorical_accuracy: 0.4509\n",
            "Epoch 35/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1249 - sparse_categorical_accuracy: 0.4533\n",
            "Epoch 36/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.1081 - sparse_categorical_accuracy: 0.4581\n",
            "Epoch 37/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.0960 - sparse_categorical_accuracy: 0.4593\n",
            "Epoch 38/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.0831 - sparse_categorical_accuracy: 0.4643\n",
            "Epoch 39/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.0717 - sparse_categorical_accuracy: 0.4632\n",
            "Epoch 40/50\n",
            "1563/1563 [==============================] - 5s 4ms/step - loss: 2.0585 - sparse_categorical_accuracy: 0.4681\n",
            "Epoch 41/50\n",
            "1563/1563 [==============================] - 6s 4ms/step - loss: 2.0453 - sparse_categorical_accuracy: 0.4734\n",
            "Epoch 42/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.0316 - sparse_categorical_accuracy: 0.4738\n",
            "Epoch 43/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.0219 - sparse_categorical_accuracy: 0.4756\n",
            "Epoch 44/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.0112 - sparse_categorical_accuracy: 0.4768\n",
            "Epoch 45/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9965 - sparse_categorical_accuracy: 0.4800\n",
            "Epoch 46/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9876 - sparse_categorical_accuracy: 0.4844\n",
            "Epoch 47/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9778 - sparse_categorical_accuracy: 0.4846\n",
            "Epoch 48/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9653 - sparse_categorical_accuracy: 0.4878\n",
            "Epoch 49/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9578 - sparse_categorical_accuracy: 0.4890\n",
            "Epoch 50/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 1.9461 - sparse_categorical_accuracy: 0.4935\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model2.evaluate(X_testing,label_fine_T)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AwQnZ7GmauTh",
        "outputId": "61b5cf78-5603-4da8-ecec-97c88a945379"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 3.6710 - sparse_categorical_accuracy: 0.2357\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[3.671020746231079, 0.23569999635219574]"
            ]
          },
          "metadata": {},
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "==>  Built ANN model using extracted features for 20 super classses. \n",
        "For this model training accuracy is 34.45 %\n",
        "For this model testing accuracy is 32.85 %"
      ],
      "metadata": {
        "id": "7UKjytMK5n87"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3 = tf.keras.Sequential(layers=[\n",
        "        tf.keras.layers.Dense(324,activation='relu',),\n",
        "        tf.keras.layers.Dense(20,activation=tf.nn.softmax)])\n",
        "\n",
        "model3.compile(optimizer=tf.keras.optimizers.SGD(), \n",
        "              loss=tf.keras.losses.SparseCategoricalCrossentropy(), \n",
        "              metrics=tf.keras.metrics.SparseCategoricalAccuracy())\n",
        "\n",
        "history = model3.fit(X_training,label_coarse, epochs=50) "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aRWjXzAqVRoz",
        "outputId": "b6c1c8a5-6869-4c35-80d1-7184bdb8f9f4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.9394 - sparse_categorical_accuracy: 0.1172\n",
            "Epoch 2/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.8112 - sparse_categorical_accuracy: 0.1810\n",
            "Epoch 3/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.6977 - sparse_categorical_accuracy: 0.2071\n",
            "Epoch 4/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.6119 - sparse_categorical_accuracy: 0.2254\n",
            "Epoch 5/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.5496 - sparse_categorical_accuracy: 0.2393\n",
            "Epoch 6/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.5038 - sparse_categorical_accuracy: 0.2479\n",
            "Epoch 7/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.4697 - sparse_categorical_accuracy: 0.2544\n",
            "Epoch 8/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.4430 - sparse_categorical_accuracy: 0.2596\n",
            "Epoch 9/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.4207 - sparse_categorical_accuracy: 0.2664\n",
            "Epoch 10/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.4021 - sparse_categorical_accuracy: 0.2710\n",
            "Epoch 11/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3860 - sparse_categorical_accuracy: 0.2752\n",
            "Epoch 12/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3712 - sparse_categorical_accuracy: 0.2790\n",
            "Epoch 13/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3579 - sparse_categorical_accuracy: 0.2832\n",
            "Epoch 14/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3464 - sparse_categorical_accuracy: 0.2867\n",
            "Epoch 15/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3353 - sparse_categorical_accuracy: 0.2909\n",
            "Epoch 16/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3250 - sparse_categorical_accuracy: 0.2935\n",
            "Epoch 17/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.3158 - sparse_categorical_accuracy: 0.2954\n",
            "Epoch 18/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.3069 - sparse_categorical_accuracy: 0.2989\n",
            "Epoch 19/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2982 - sparse_categorical_accuracy: 0.3006\n",
            "Epoch 20/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2905 - sparse_categorical_accuracy: 0.3034\n",
            "Epoch 21/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2826 - sparse_categorical_accuracy: 0.3044\n",
            "Epoch 22/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2755 - sparse_categorical_accuracy: 0.3067\n",
            "Epoch 23/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2679 - sparse_categorical_accuracy: 0.3091\n",
            "Epoch 24/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2612 - sparse_categorical_accuracy: 0.3109\n",
            "Epoch 25/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.2549 - sparse_categorical_accuracy: 0.3120\n",
            "Epoch 26/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2483 - sparse_categorical_accuracy: 0.3136\n",
            "Epoch 27/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2420 - sparse_categorical_accuracy: 0.3150\n",
            "Epoch 28/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2363 - sparse_categorical_accuracy: 0.3170\n",
            "Epoch 29/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2301 - sparse_categorical_accuracy: 0.3188\n",
            "Epoch 30/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2246 - sparse_categorical_accuracy: 0.3213\n",
            "Epoch 31/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2192 - sparse_categorical_accuracy: 0.3223\n",
            "Epoch 32/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2132 - sparse_categorical_accuracy: 0.3234\n",
            "Epoch 33/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2079 - sparse_categorical_accuracy: 0.3249\n",
            "Epoch 34/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.2024 - sparse_categorical_accuracy: 0.3253\n",
            "Epoch 35/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.1975 - sparse_categorical_accuracy: 0.3266\n",
            "Epoch 36/50\n",
            "1563/1563 [==============================] - 5s 3ms/step - loss: 2.1924 - sparse_categorical_accuracy: 0.3287\n",
            "Epoch 37/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1875 - sparse_categorical_accuracy: 0.3307\n",
            "Epoch 38/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1824 - sparse_categorical_accuracy: 0.3314\n",
            "Epoch 39/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1778 - sparse_categorical_accuracy: 0.3328\n",
            "Epoch 40/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1728 - sparse_categorical_accuracy: 0.3324\n",
            "Epoch 41/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1683 - sparse_categorical_accuracy: 0.3346\n",
            "Epoch 42/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1638 - sparse_categorical_accuracy: 0.3361\n",
            "Epoch 43/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1595 - sparse_categorical_accuracy: 0.3379\n",
            "Epoch 44/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1553 - sparse_categorical_accuracy: 0.3389\n",
            "Epoch 45/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1505 - sparse_categorical_accuracy: 0.3410\n",
            "Epoch 46/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1461 - sparse_categorical_accuracy: 0.3413\n",
            "Epoch 47/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1421 - sparse_categorical_accuracy: 0.3429\n",
            "Epoch 48/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1377 - sparse_categorical_accuracy: 0.3442\n",
            "Epoch 49/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1336 - sparse_categorical_accuracy: 0.3459\n",
            "Epoch 50/50\n",
            "1563/1563 [==============================] - 4s 3ms/step - loss: 2.1294 - sparse_categorical_accuracy: 0.3445\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model3.evaluate(X_testing,label_coarse_T)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "I5l2UFDbWIBv",
        "outputId": "deacafa1-03ee-437c-d54d-eba55dd53d69"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 2.1873 - sparse_categorical_accuracy: 0.3286\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[2.1873350143432617, 0.3285999894142151]"
            ]
          },
          "metadata": {},
          "execution_count": 54
        }
      ]
    }
  ]
}