Denoising_autoencoder.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Denoising autoencoder.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNvJsj46OsOAquoeLnwcsdK",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/huytranvan2010/huytranvan2010.github.io/blob/main/Denoising_autoencoder.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 129
        },
        "id": "Ax7JuCfZugJI",
        "outputId": "d31e0b03-7408-44f5-b88e-b97a02795f85"
      },
      "source": [
        "from tensorflow.keras.datasets import mnist\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras import Model\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "(x_train, _), (x_test, _) = mnist.load_data()\n",
        "\n",
        "# normalize và reshape\n",
        "x_train = x_train.astype('float32') / 255.\n",
        "x_test = x_test.astype('float32') / 255.\n",
        "x_train = np.reshape(x_train, (len(x_train), 28, 28, 1))\n",
        "x_test = np.reshape(x_test, (len(x_test), 28, 28, 1))\n",
        "\n",
        "# add noise vào ảnh\n",
        "noise_factor = 0.5\n",
        "x_train_noisy = x_train + noise_factor * np.random.normal(loc=0.0, scale=1.0, size=x_train.shape)   # size đúng với x_train để kớp\n",
        "x_test_noisy = x_test + noise_factor * np.random.normal(loc=0.0, scale=1.0, size=x_test.shape) \n",
        "\n",
        "# clip, giá trị nào lớn hơn 1 cho về 1, nhỏ hơn 0 cho về 0\n",
        "x_train_noisy = np.clip(x_train_noisy, 0., 1.)\n",
        "x_test_noisy = np.clip(x_test_noisy, 0., 1.)\n",
        "\n",
        "# hiển thị một số ảnh\n",
        "n = 10\n",
        "plt.figure(figsize=(20,2))\n",
        "for i in range(1, n + 1):\n",
        "    ax = plt.subplot(1, n, i)\n",
        "    plt.imshow(x_test_noisy[i].reshape(28, 28))     # ảnh bị normalize về [0, 1] mà vẫn hiển thị nhỉ\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n",
        "plt.show()\n"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x144 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f1x7urKRv-ix",
        "outputId": "779e8ee1-8983-4247-b2fd-b611b67fabf2"
      },
      "source": [
        "# Xay dung model\n",
        "input_img = layers.Input(shape=(28, 28, 1))\n",
        "\n",
        "x = layers.Conv2D(32, (3, 3), activation='relu', padding='same')(input_img)\n",
        "x = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
        "x = layers.Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
        "encoded = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
        "\n",
        "# At this point the representation is (7, 7, 32)\n",
        "# Decoder y hệt như encoder nhưng theo hướng ngược lại\n",
        "# Có thể dùng Conv2DTranspose hoặc Upsampling2D\n",
        "\n",
        "x = layers.Conv2D(32, (3, 3), activation='relu', padding='same')(encoded)\n",
        "x = layers.UpSampling2D((2, 2))(x)\n",
        "x = layers.Conv2D(32, (3, 3), activation='relu', padding='same')(x)\n",
        "x = layers.UpSampling2D((2, 2))(x)\n",
        "decoded = layers.Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)\n",
        "\n",
        "autoencoder = Model(inputs=input_img, outputs=decoded)\n",
        "autoencoder.compile(optimizer='adam', loss='binary_crossentropy')\n",
        "\n",
        "autoencoder.summary()"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_4\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_5 (InputLayer)         [(None, 28, 28, 1)]       0         \n",
            "_________________________________________________________________\n",
            "conv2d_20 (Conv2D)           (None, 28, 28, 32)        320       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_8 (MaxPooling2 (None, 14, 14, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_21 (Conv2D)           (None, 14, 14, 32)        9248      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_9 (MaxPooling2 (None, 7, 7, 32)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_22 (Conv2D)           (None, 7, 7, 32)          9248      \n",
            "_________________________________________________________________\n",
            "up_sampling2d_8 (UpSampling2 (None, 14, 14, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_23 (Conv2D)           (None, 14, 14, 32)        9248      \n",
            "_________________________________________________________________\n",
            "up_sampling2d_9 (UpSampling2 (None, 28, 28, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_24 (Conv2D)           (None, 28, 28, 1)         289       \n",
            "=================================================================\n",
            "Total params: 28,353\n",
            "Trainable params: 28,353\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IwTPzxHnwORR",
        "outputId": "f65ba6ae-68e2-464f-8d1c-02659814a66b"
      },
      "source": [
        "NUM_EPOCHS = 20\n",
        "BATCH_SIZE = 128\n",
        "\n",
        "hist = autoencoder.fit(x_train_noisy, x_train,\n",
        "                epochs=NUM_EPOCHS,\n",
        "                batch_size=BATCH_SIZE,\n",
        "                shuffle=True,\n",
        "                validation_data=(x_test_noisy, x_test),\n",
        "                callbacks=[tf.keras.callbacks.TensorBoard(log_dir='/tmp/tb', histogram_freq=0, write_graph=False)])"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "469/469 [==============================] - 4s 8ms/step - loss: 0.0953 - val_loss: 0.0949\n",
            "Epoch 2/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0952 - val_loss: 0.0954\n",
            "Epoch 3/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0951 - val_loss: 0.0947\n",
            "Epoch 4/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0950 - val_loss: 0.0947\n",
            "Epoch 5/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0949 - val_loss: 0.0945\n",
            "Epoch 6/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0948 - val_loss: 0.0945\n",
            "Epoch 7/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0947 - val_loss: 0.0943\n",
            "Epoch 8/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0945 - val_loss: 0.0944\n",
            "Epoch 9/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0945 - val_loss: 0.0942\n",
            "Epoch 10/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0944 - val_loss: 0.0940\n",
            "Epoch 11/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0944 - val_loss: 0.0942\n",
            "Epoch 12/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0943 - val_loss: 0.0940\n",
            "Epoch 13/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0942 - val_loss: 0.0941\n",
            "Epoch 14/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0942 - val_loss: 0.0941\n",
            "Epoch 15/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0941 - val_loss: 0.0938\n",
            "Epoch 16/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0941 - val_loss: 0.0938\n",
            "Epoch 17/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0940 - val_loss: 0.0943\n",
            "Epoch 18/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0940 - val_loss: 0.0938\n",
            "Epoch 19/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0940 - val_loss: 0.0938\n",
            "Epoch 20/20\n",
            "469/469 [==============================] - 3s 7ms/step - loss: 0.0939 - val_loss: 0.0937\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "2ga6I6FUvUX-",
        "outputId": "cf505498-51df-4227-c086-a95886e9e543"
      },
      "source": [
        "# funstion for train and val losses visualizations\n",
        "def plot_losses(hist):\n",
        "    plt.plot(np.arange(NUM_EPOCHS), hist.history['loss'], label='Training loss')\n",
        "    plt.plot(np.arange(NUM_EPOCHS), hist.history['val_loss'],label='Validation loss')\n",
        "    plt.title('Training and validation loss')\n",
        "    plt.xlabel('Epochs')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    plt.close()\n",
        "\n",
        "# visualize train and val losses\n",
        "plot_losses(hist)\n"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WNNa8E3f4P7r"
      },
      "source": [
        "## Kiểm tra lại"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "wSFlLPf7yvlf",
        "outputId": "97c2028f-af80-495b-bff4-61fd23bcbc6a"
      },
      "source": [
        "import numpy as np\n",
        "\n",
        "img = x_test_noisy[5]\n",
        "img = img[np.newaxis, ...]      # đúng shape(1, 28, 28, 1)\n",
        "\n",
        "print(img.shape)\n",
        "new_img = autoencoder.predict(img)\n",
        "\n",
        "# post processing\n",
        "new_img = new_img * 255\n",
        "new_img = new_img.astype(\"uint8\")   # trả về unsigned integer\n",
        "\n",
        "plt.imshow(new_img.reshape(28, 28))     # reshape lại. Cái này để chính xác nên dùng tf.squeeze để dropout các axis cho chính xác\n",
        "plt.show()"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1, 28, 28, 1)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OkOPYWy3yusL"
      },
      "source": [
        ""
      ]
    }
  ]
}