UNet/UNet_Biomedical_segmentation.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "UNet Biomedical segmentation.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNgtVoIa5je530dWzec4Sg8",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/farazkhanfk7/dl-models/blob/master/UNet/UNet_Biomedical_segmentation.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vKVMZquo4-BA",
        "colab_type": "text"
      },
      "source": [
        "# U-Net Architecture"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f3wBbimJ6W1M",
        "colab_type": "text"
      },
      "source": [
        "![picture](https://img.techpowerup.org/200813/u-net-architecture.png)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WcT5ciCEtKmR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Data download and unzipping\n",
        "!wget https://raw.githubusercontent.com/AakashSudhakar/2018-data-science-bowl/master/compressed_files/stage1_test.zip -c\n",
        "!wget https://raw.githubusercontent.com/AakashSudhakar/2018-data-science-bowl/master/compressed_files/stage1_train.zip -c\n",
        "\n",
        "!mkdir stage1_train stage1_test\n",
        "\n",
        "!unzip stage1_train.zip -d stage1_train/\n",
        "!unzip stage1_test.zip -d stage1_test/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hOHFFZMotTLa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os\n",
        "os.chdir('/content')"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-X2MX4oDt9EN",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "8d488eb7-1990-438f-be9e-5c8df8c22850"
      },
      "source": [
        "!pwd"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qd6-1nOhtypu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "4e8ae1a0-3d47-47a1-b62b-5fe2b4209e35"
      },
      "source": [
        "!ls"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "sample_data  stage1_test  stage1_test.zip  stage1_train  stage1_train.zip\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SMgTdLa8tzxT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## Imports\n",
        "import os\n",
        "import sys\n",
        "import random\n",
        "\n",
        "import numpy as np\n",
        "import cv2\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "\n",
        "## Seeding \n",
        "seed = 2019\n",
        "random.seed = seed\n",
        "np.random.seed = seed\n",
        "tf.seed = seed"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "koMHeBLX7CwR",
        "colab_type": "text"
      },
      "source": [
        "# Data-Generator"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PgayXf9Nxj5S",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class DataGen(keras.utils.Sequence):\n",
        "    def __init__(self, ids, path, batch_size=8, image_size=128):\n",
        "        self.ids = ids\n",
        "        self.path = path\n",
        "        self.batch_size = batch_size\n",
        "        self.image_size = image_size\n",
        "        self.on_epoch_end()\n",
        "        \n",
        "    def __load__(self, id_name):\n",
        "        ## Path\n",
        "        image_path = os.path.join(self.path, id_name, \"images\", id_name) + \".png\"\n",
        "        mask_path = os.path.join(self.path, id_name, \"masks/\")\n",
        "        all_masks = os.listdir(mask_path)\n",
        "        \n",
        "        ## Reading Image\n",
        "        image = cv2.imread(image_path, 1)\n",
        "        image = cv2.resize(image, (self.image_size, self.image_size))\n",
        "        \n",
        "        mask = np.zeros((self.image_size, self.image_size, 1))\n",
        "        \n",
        "        ## Reading Masks\n",
        "        for name in all_masks:\n",
        "            _mask_path = mask_path + name\n",
        "            _mask_image = cv2.imread(_mask_path, -1)\n",
        "            _mask_image = cv2.resize(_mask_image, (self.image_size, self.image_size)) #128x128\n",
        "            _mask_image = np.expand_dims(_mask_image, axis=-1)\n",
        "            mask = np.maximum(mask, _mask_image)\n",
        "            \n",
        "        ## Normalizaing \n",
        "        image = image/255.0\n",
        "        mask = mask/255.0\n",
        "        \n",
        "        return image, mask\n",
        "    \n",
        "    def __getitem__(self, index):\n",
        "        if(index+1)*self.batch_size > len(self.ids):\n",
        "            self.batch_size = len(self.ids) - index*self.batch_size\n",
        "        \n",
        "        files_batch = self.ids[index*self.batch_size : (index+1)*self.batch_size]\n",
        "        \n",
        "        image = []\n",
        "        mask  = []\n",
        "        \n",
        "        for id_name in files_batch:\n",
        "            _img, _mask = self.__load__(id_name)\n",
        "            image.append(_img)\n",
        "            mask.append(_mask)\n",
        "            \n",
        "        image = np.array(image)\n",
        "        mask  = np.array(mask)\n",
        "        \n",
        "        return image, mask\n",
        "    \n",
        "    def on_epoch_end(self):\n",
        "        pass\n",
        "    \n",
        "    def __len__(self):\n",
        "        return int(np.ceil(len(self.ids)/float(self.batch_size)))"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IbxSEsAr7pro",
        "colab_type": "text"
      },
      "source": [
        "# Hyper-Parameter"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WA3Im2kSxyN5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "image_size = 128\n",
        "train_path = \"stage1_train\"\n",
        "epochs = 5\n",
        "batch_size = 8\n",
        "\n",
        "## Training Ids\n",
        "train_ids = next(os.walk(train_path))[1]\n",
        "\n",
        "## Validation Data Size\n",
        "val_data_size = 10\n",
        "\n",
        "valid_ids = train_ids[:val_data_size]\n",
        "train_ids = train_ids[val_data_size:]"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bNfb5snf3FBM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "76bf2c7e-e2c0-47df-e5eb-7c8479c1ecdd"
      },
      "source": [
        "len(train_ids)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "660"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jqots9pa1SWm",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "0d42162b-875a-478d-92c1-98d6f94a157e"
      },
      "source": [
        "gen = DataGen(train_ids, train_path, batch_size=batch_size, image_size=image_size)\n",
        "x, y = gen.__getitem__(0)\n",
        "print(x.shape, y.shape)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8, 128, 128, 3) (8, 128, 128, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bq4Wt8EC1VaP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 208
        },
        "outputId": "a3d7ec6b-9d5b-4a71-fc3c-d40b27e9ec9c"
      },
      "source": [
        "r = random.randint(0, len(x)-1)\n",
        "\n",
        "fig = plt.figure()\n",
        "fig.subplots_adjust(hspace=0.4, wspace=0.4)\n",
        "ax = fig.add_subplot(1, 2, 1)\n",
        "ax.imshow(x[r])\n",
        "ax = fig.add_subplot(1, 2, 2)\n",
        "ax.imshow(np.reshape(y[r], (image_size, image_size)), cmap=\"viridis\")"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f8d1de3b2b0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EBlVumtS7vZf",
        "colab_type": "text"
      },
      "source": [
        "# Different Convolutional Blocks"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yTrS80jh1Yvv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def down_block(x, filters, kernel_size=(3, 3), padding=\"same\", strides=1):\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(x)\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(c)\n",
        "    p = keras.layers.MaxPool2D((2, 2), (2, 2))(c)\n",
        "    return c, p\n",
        "\n",
        "def up_block(x, skip, filters, kernel_size=(3, 3), padding=\"same\", strides=1):\n",
        "    us = keras.layers.UpSampling2D((2, 2))(x)\n",
        "    concat = keras.layers.Concatenate()([us, skip])\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(concat)\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(c)\n",
        "    return c\n",
        "\n",
        "def bottleneck(x, filters, kernel_size=(3, 3), padding=\"same\", strides=1):\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(x)\n",
        "    c = keras.layers.Conv2D(filters, kernel_size, padding=padding, strides=strides, activation=\"relu\")(c)\n",
        "    return c"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z9GpcULg70kv",
        "colab_type": "text"
      },
      "source": [
        "# Building the UNet model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1gurWvla2QXc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def UNet():\n",
        "    f = [16, 32, 64, 128, 256]\n",
        "    inputs = keras.layers.Input((image_size, image_size, 3))\n",
        "    \n",
        "    p0 = inputs\n",
        "    c1, p1 = down_block(p0, f[0]) #128 -> 64\n",
        "    c2, p2 = down_block(p1, f[1]) #64 -> 32\n",
        "    c3, p3 = down_block(p2, f[2]) #32 -> 16\n",
        "    c4, p4 = down_block(p3, f[3]) #16->8\n",
        "    \n",
        "    bn = bottleneck(p4, f[4])\n",
        "    \n",
        "    u1 = up_block(bn, c4, f[3]) #8 -> 16\n",
        "    u2 = up_block(u1, c3, f[2]) #16 -> 32\n",
        "    u3 = up_block(u2, c2, f[1]) #32 -> 64\n",
        "    u4 = up_block(u3, c1, f[0]) #64 -> 128\n",
        "    \n",
        "    outputs = keras.layers.Conv2D(1, (1, 1), padding=\"same\", activation=\"sigmoid\")(u4)\n",
        "    model = keras.models.Model(inputs, outputs)\n",
        "    return model"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3ZWLCRU42Sh8",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "5ba6a989-3008-4fdd-d32f-353a6badc6df"
      },
      "source": [
        "model = UNet()\n",
        "model.compile(optimizer=\"adam\", loss=\"binary_crossentropy\", metrics=[\"acc\"])\n",
        "model.summary()"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_1\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 128, 128, 3) 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d (Conv2D)                 (None, 128, 128, 16) 448         input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_1 (Conv2D)               (None, 128, 128, 16) 2320        conv2d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D)    (None, 64, 64, 16)   0           conv2d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_2 (Conv2D)               (None, 64, 64, 32)   4640        max_pooling2d[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_3 (Conv2D)               (None, 64, 64, 32)   9248        conv2d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2D)  (None, 32, 32, 32)   0           conv2d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_4 (Conv2D)               (None, 32, 32, 64)   18496       max_pooling2d_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_5 (Conv2D)               (None, 32, 32, 64)   36928       conv2d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2D)  (None, 16, 16, 64)   0           conv2d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_6 (Conv2D)               (None, 16, 16, 128)  73856       max_pooling2d_2[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_7 (Conv2D)               (None, 16, 16, 128)  147584      conv2d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2D)  (None, 8, 8, 128)    0           conv2d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_8 (Conv2D)               (None, 8, 8, 256)    295168      max_pooling2d_3[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_9 (Conv2D)               (None, 8, 8, 256)    590080      conv2d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d (UpSampling2D)    (None, 16, 16, 256)  0           conv2d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 16, 16, 384)  0           up_sampling2d[0][0]              \n",
            "                                                                 conv2d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_10 (Conv2D)              (None, 16, 16, 128)  442496      concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_11 (Conv2D)              (None, 16, 16, 128)  147584      conv2d_10[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_1 (UpSampling2D)  (None, 32, 32, 128)  0           conv2d_11[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_1 (Concatenate)     (None, 32, 32, 192)  0           up_sampling2d_1[0][0]            \n",
            "                                                                 conv2d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_12 (Conv2D)              (None, 32, 32, 64)   110656      concatenate_1[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_13 (Conv2D)              (None, 32, 32, 64)   36928       conv2d_12[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_2 (UpSampling2D)  (None, 64, 64, 64)   0           conv2d_13[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_2 (Concatenate)     (None, 64, 64, 96)   0           up_sampling2d_2[0][0]            \n",
            "                                                                 conv2d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_14 (Conv2D)              (None, 64, 64, 32)   27680       concatenate_2[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_15 (Conv2D)              (None, 64, 64, 32)   9248        conv2d_14[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_3 (UpSampling2D)  (None, 128, 128, 32) 0           conv2d_15[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_3 (Concatenate)     (None, 128, 128, 48) 0           up_sampling2d_3[0][0]            \n",
            "                                                                 conv2d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_16 (Conv2D)              (None, 128, 128, 16) 6928        concatenate_3[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_17 (Conv2D)              (None, 128, 128, 16) 2320        conv2d_16[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_18 (Conv2D)              (None, 128, 128, 1)  17          conv2d_17[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 1,962,625\n",
            "Trainable params: 1,962,625\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2glQpGAZ76J3",
        "colab_type": "text"
      },
      "source": [
        "# Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XcXoIYS42UfL",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 275
        },
        "outputId": "dd81250c-8a0b-46dc-8b50-3e74411f62e5"
      },
      "source": [
        "train_gen = DataGen(train_ids, train_path, image_size=image_size, batch_size=batch_size)\n",
        "valid_gen = DataGen(valid_ids, train_path, image_size=image_size, batch_size=batch_size)\n",
        "\n",
        "train_steps = len(train_ids)//batch_size\n",
        "valid_steps = len(valid_ids)//batch_size\n",
        "\n",
        "model.fit_generator(train_gen, validation_data=valid_gen, steps_per_epoch=train_steps, validation_steps=valid_steps, \n",
        "                    epochs=epochs)"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-27-36281b7dec54>:8: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use Model.fit, which supports generators.\n",
            "Epoch 1/5\n",
            "82/82 [==============================] - 24s 290ms/step - loss: 0.5709 - acc: 0.8454 - val_loss: 0.2458 - val_acc: 0.9052\n",
            "Epoch 2/5\n",
            "82/82 [==============================] - 12s 144ms/step - loss: 0.2223 - acc: 0.8877 - val_loss: 0.1741 - val_acc: 0.9180\n",
            "Epoch 3/5\n",
            "82/82 [==============================] - 11s 139ms/step - loss: 0.1390 - acc: 0.9310 - val_loss: 0.0786 - val_acc: 0.9678\n",
            "Epoch 4/5\n",
            "82/82 [==============================] - 13s 157ms/step - loss: 0.1114 - acc: 0.9413 - val_loss: 0.0448 - val_acc: 0.9730\n",
            "Epoch 5/5\n",
            "82/82 [==============================] - 12s 149ms/step - loss: 0.0954 - acc: 0.9488 - val_loss: 0.0418 - val_acc: 0.9731\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f8d10179898>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "siewk8bW2d1M",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## Save the Weights\n",
        "model.save_weights(\"UNetW.h5\")\n",
        "\n",
        "## Dataset for prediction\n",
        "x, y = valid_gen.__getitem__(1)\n",
        "result = model.predict(x)\n",
        "\n",
        "result = result > 0.5"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "33Q9BbZs3b9R",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 208
        },
        "outputId": "20f68575-0440-42bd-e865-01763834523f"
      },
      "source": [
        "fig = plt.figure()\n",
        "fig.subplots_adjust(hspace=0.4, wspace=0.4)\n",
        "\n",
        "ax = fig.add_subplot(1, 2, 1)\n",
        "ax.imshow(np.reshape(y[0]*255, (image_size, image_size)), cmap=\"gray\")\n",
        "\n",
        "ax = fig.add_subplot(1, 2, 2)\n",
        "ax.imshow(np.reshape(result[0]*255, (image_size, image_size)), cmap=\"viridis\")"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f8cba5e9ba8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}