Practice-3_Fashion_MNIST_Computer_Vision_Accuracy_using_Convolutions.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Improving_Computer_Vision_Accuracy_using_Convolutions.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "J_LQ0UeVwCf6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 612
        },
        "outputId": "2ad83506-e78c-4f6b-bec8-50cec144ef1e"
      },
      "source": [
        "import tensorflow as tf\n",
        "mnist = tf.keras.datasets.fashion_mnist\n",
        "(training_images, training_labels),(test_images, test_labels) = mnist.load_data()\n",
        "training_images = training_images.reshape(60000, 28, 28, 1)\n",
        "training_images = training_images/255.0\n",
        "test_images = test_images.reshape(10000, 28,28,1)\n",
        "test_images = test_images/255.0\n",
        "model = tf.keras.models.Sequential([tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(28,28,1)),\n",
        "                                    tf.keras.layers.MaxPooling2D(2,2),\n",
        "                                    tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n",
        "                                    tf.keras.layers.MaxPooling2D(2,2),\n",
        "                                    tf.keras.layers.Flatten(),\n",
        "                                    tf.keras.layers.Dense(128, activation=tf.nn.relu),\n",
        "                                    tf.keras.layers.Dense(10, activation=tf.nn.softmax)])\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "model.summary()\n",
        "model.fit(training_images, training_labels, epochs=5)\n",
        "test_loss, test_acc = model.evaluate(test_images, test_labels)\n",
        "print(test_loss)\n",
        "print(test_acc)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_11\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_7 (Conv2D)            (None, 26, 26, 64)        640       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_7 (MaxPooling2 (None, 13, 13, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 11, 11, 64)        36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_8 (MaxPooling2 (None, 5, 5, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten_11 (Flatten)         (None, 1600)              0         \n",
            "_________________________________________________________________\n",
            "dense_22 (Dense)             (None, 128)               204928    \n",
            "_________________________________________________________________\n",
            "dense_23 (Dense)             (None, 10)                1290      \n",
            "=================================================================\n",
            "Total params: 243,786\n",
            "Trainable params: 243,786\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/5\n",
            "1875/1875 [==============================] - 81s 43ms/step - loss: 0.4432 - accuracy: 0.8403\n",
            "Epoch 2/5\n",
            "1875/1875 [==============================] - 84s 45ms/step - loss: 0.3006 - accuracy: 0.8895\n",
            "Epoch 3/5\n",
            "1875/1875 [==============================] - 80s 43ms/step - loss: 0.2575 - accuracy: 0.9052\n",
            "Epoch 4/5\n",
            "1875/1875 [==============================] - 80s 43ms/step - loss: 0.2264 - accuracy: 0.9159\n",
            "Epoch 5/5\n",
            "1875/1875 [==============================] - 82s 44ms/step - loss: 0.1990 - accuracy: 0.9261\n",
            "313/313 [==============================] - 4s 14ms/step - loss: 0.2530 - accuracy: 0.9137\n",
            "0.25298798084259033\n",
            "0.9136999845504761\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4z4HzzUxjGKq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "outputId": "e75812c6-bc39-4ebd-fc06-db4e1e9a4ddd"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "f, axarr = plt.subplots(3,4)\n",
        "FIRST_IMAGE=0\n",
        "SECOND_IMAGE=7\n",
        "THIRD_IMAGE=26\n",
        "CONVOLUTION_NUMBER = 1\n",
        "from tensorflow.keras import models\n",
        "layer_outputs = [layer.output for layer in model.layers]\n",
        "activation_model = tf.keras.models.Model(inputs = model.input, outputs = layer_outputs)\n",
        "for x in range(0,4):\n",
        "  f1 = activation_model.predict(test_images[FIRST_IMAGE].reshape(1, 28, 28, 1))[x]\n",
        "  axarr[0,x].imshow(f1[0, :, :, CONVOLUTION_NUMBER], cmap='inferno')\n",
        "  axarr[0,x].grid(False)\n",
        "  f2 = activation_model.predict(test_images[SECOND_IMAGE].reshape(1,28,28,1))[x]\n",
        "  axarr[1,x].imshow(f2[0, :, :,CONVOLUTION_NUMBER], cmap='inferno')\n",
        "  axarr[1,x].grid(False)\n",
        "  f3 = activation_model.predict(test_images[THIRD_IMAGE].reshape(1,28,28,1))[x]\n",
        "  axarr[2,x].imshow(f3[0,: , : ,CONVOLUTION_NUMBER], cmap='inferno')\n",
        "  axarr[2,x].grid(False)"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 12 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yJEpIngKfLWo",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 714
        },
        "outputId": "da244173-ea49-487e-f0c1-fea84c8a59d5"
      },
      "source": [
        "import tensorflow as tf\n",
        "mnist = tf.keras.datasets.fashion_mnist\n",
        "(training_images, training_labels),(test_images, test_labels) = mnist.load_data()\n",
        "training_images = training_images.reshape(60000, 28, 28, 1)\n",
        "training_images = training_images/255.0\n",
        "test_images = test_images.reshape(10000, 28,28,1)\n",
        "test_images = test_images/255.0\n",
        "model = tf.keras.models.Sequential([tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(28,28,1)),\n",
        "                                    tf.keras.layers.MaxPooling2D(2,2),\n",
        "                                    tf.keras.layers.Flatten(),\n",
        "                                    tf.keras.layers.Dense(128, activation=tf.nn.relu),\n",
        "                                    tf.keras.layers.Dense(10, activation=tf.nn.softmax)])\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "model.summary()\n",
        "model.fit(training_images, training_labels, epochs=10)\n",
        "test_loss, test_acc = model.evaluate(test_images, test_labels)\n",
        "print(test_loss)\n",
        "print(test_acc)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_14\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_11 (Conv2D)           (None, 26, 26, 64)        640       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_11 (MaxPooling (None, 13, 13, 64)        0         \n",
            "_________________________________________________________________\n",
            "flatten_14 (Flatten)         (None, 10816)             0         \n",
            "_________________________________________________________________\n",
            "dense_28 (Dense)             (None, 128)               1384576   \n",
            "_________________________________________________________________\n",
            "dense_29 (Dense)             (None, 10)                1290      \n",
            "=================================================================\n",
            "Total params: 1,386,506\n",
            "Trainable params: 1,386,506\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Epoch 1/10\n",
            "1875/1875 [==============================] - 54s 29ms/step - loss: 0.3743 - accuracy: 0.8679\n",
            "Epoch 2/10\n",
            "1875/1875 [==============================] - 54s 29ms/step - loss: 0.2513 - accuracy: 0.9080\n",
            "Epoch 3/10\n",
            "1875/1875 [==============================] - 53s 28ms/step - loss: 0.2067 - accuracy: 0.9230\n",
            "Epoch 4/10\n",
            "1875/1875 [==============================] - 53s 28ms/step - loss: 0.1701 - accuracy: 0.9369\n",
            "Epoch 5/10\n",
            "1875/1875 [==============================] - 53s 28ms/step - loss: 0.1436 - accuracy: 0.9468\n",
            "Epoch 6/10\n",
            "1875/1875 [==============================] - 54s 29ms/step - loss: 0.1203 - accuracy: 0.9549\n",
            "Epoch 7/10\n",
            "1875/1875 [==============================] - 53s 28ms/step - loss: 0.0976 - accuracy: 0.9632\n",
            "Epoch 8/10\n",
            "1875/1875 [==============================] - 54s 29ms/step - loss: 0.0803 - accuracy: 0.9708\n",
            "Epoch 9/10\n",
            "1875/1875 [==============================] - 57s 30ms/step - loss: 0.0670 - accuracy: 0.9763\n",
            "Epoch 10/10\n",
            "1875/1875 [==============================] - 53s 28ms/step - loss: 0.0548 - accuracy: 0.9808\n",
            "313/313 [==============================] - 3s 9ms/step - loss: 0.3517 - accuracy: 0.9162\n",
            "0.3517310917377472\n",
            "0.9161999821662903\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RgAsvdK1EGkW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}