Depth_Data_2D_Keras.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y6dWPdToXniE"
      },
      "source": [
        "Get depth data from 2D Images.\n",
        "\n",
        "Created by - Achint Sharma"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Imports and setup"
      ],
      "metadata": {
        "id": "neyMa0J-Th29"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip3 install pixellib"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gs82KbpSNkR0",
        "outputId": "0a65b553-71be-42b9-80fd-7c4a0f25bce3"
      },
      "execution_count": 139,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: pixellib in /usr/local/lib/python3.7/dist-packages (0.7.1)\n",
            "Requirement already satisfied: labelme2coco in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.2.1)\n",
            "Requirement already satisfied: pillow in /usr/local/lib/python3.7/dist-packages (from pixellib) (9.1.0)\n",
            "Requirement already satisfied: cloudpickle in /usr/local/lib/python3.7/dist-packages (from pixellib) (1.3.0)\n",
            "Requirement already satisfied: opencv-python in /usr/local/lib/python3.7/dist-packages (from pixellib) (4.5.5.64)\n",
            "Requirement already satisfied: fvcore in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.1.5.post20220504)\n",
            "Requirement already satisfied: omegaconf in /usr/local/lib/python3.7/dist-packages (from pixellib) (2.1.2)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from pixellib) (4.64.0)\n",
            "Requirement already satisfied: imantics in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.1.12)\n",
            "Requirement already satisfied: scikit-image in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.18.3)\n",
            "Requirement already satisfied: termcolor in /usr/local/lib/python3.7/dist-packages (from pixellib) (1.1.0)\n",
            "Requirement already satisfied: pydot in /usr/local/lib/python3.7/dist-packages (from pixellib) (1.3.0)\n",
            "Requirement already satisfied: hydra-core in /usr/local/lib/python3.7/dist-packages (from pixellib) (1.1.2)\n",
            "Requirement already satisfied: yacs in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.1.8)\n",
            "Requirement already satisfied: iopath in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.1.9)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (from pixellib) (3.2.2)\n",
            "Requirement already satisfied: dataclasses in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.6)\n",
            "Requirement already satisfied: pyQt5 in /usr/local/lib/python3.7/dist-packages (from pixellib) (5.15.6)\n",
            "Requirement already satisfied: future in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.16.0)\n",
            "Requirement already satisfied: Cython in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.29.28)\n",
            "Requirement already satisfied: black in /usr/local/lib/python3.7/dist-packages (from pixellib) (22.3.0)\n",
            "Requirement already satisfied: tabulate in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.8.9)\n",
            "Requirement already satisfied: imgaug in /usr/local/lib/python3.7/dist-packages (from pixellib) (0.2.9)\n",
            "Requirement already satisfied: click>=8.0.0 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (8.0.4)\n",
            "Requirement already satisfied: tomli>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (2.0.1)\n",
            "Requirement already satisfied: pathspec>=0.9.0 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (0.9.0)\n",
            "Requirement already satisfied: platformdirs>=2 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (2.5.2)\n",
            "Requirement already satisfied: typed-ast>=1.4.2 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (1.5.3)\n",
            "Requirement already satisfied: mypy-extensions>=0.4.3 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (0.4.3)\n",
            "Requirement already satisfied: typing-extensions>=3.10.0.0 in /usr/local/lib/python3.7/dist-packages (from black->pixellib) (4.2.0)\n",
            "Requirement already satisfied: importlib-metadata in /usr/local/lib/python3.7/dist-packages (from click>=8.0.0->black->pixellib) (4.11.3)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.7/dist-packages (from fvcore->pixellib) (6.0)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from fvcore->pixellib) (1.21.6)\n",
            "Requirement already satisfied: portalocker in /usr/local/lib/python3.7/dist-packages (from iopath->pixellib) (2.4.0)\n",
            "Requirement already satisfied: importlib-resources<5.3 in /usr/local/lib/python3.7/dist-packages (from hydra-core->pixellib) (5.2.3)\n",
            "Requirement already satisfied: antlr4-python3-runtime==4.8 in /usr/local/lib/python3.7/dist-packages (from hydra-core->pixellib) (4.8)\n",
            "Requirement already satisfied: zipp>=3.1.0 in /usr/local/lib/python3.7/dist-packages (from importlib-resources<5.3->hydra-core->pixellib) (3.8.0)\n",
            "Requirement already satisfied: lxml in /usr/local/lib/python3.7/dist-packages (from imantics->pixellib) (4.2.6)\n",
            "Requirement already satisfied: xmljson in /usr/local/lib/python3.7/dist-packages (from imantics->pixellib) (0.2.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from imgaug->pixellib) (1.15.0)\n",
            "Requirement already satisfied: imageio in /usr/local/lib/python3.7/dist-packages (from imgaug->pixellib) (2.4.1)\n",
            "Requirement already satisfied: Shapely in /usr/local/lib/python3.7/dist-packages (from imgaug->pixellib) (1.8.1.post1)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from imgaug->pixellib) (1.4.1)\n",
            "Requirement already satisfied: networkx>=2.0 in /usr/local/lib/python3.7/dist-packages (from scikit-image->pixellib) (2.6.3)\n",
            "Requirement already satisfied: tifffile>=2019.7.26 in /usr/local/lib/python3.7/dist-packages (from scikit-image->pixellib) (2021.11.2)\n",
            "Requirement already satisfied: PyWavelets>=1.1.1 in /usr/local/lib/python3.7/dist-packages (from scikit-image->pixellib) (1.3.0)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib->pixellib) (0.11.0)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->pixellib) (2.8.2)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->pixellib) (3.0.8)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->pixellib) (1.4.2)\n",
            "Requirement already satisfied: sahi>=0.8.19 in /usr/local/lib/python3.7/dist-packages (from labelme2coco->pixellib) (0.9.2)\n",
            "Requirement already satisfied: jsonschema>=2.6.0 in /usr/local/lib/python3.7/dist-packages (from labelme2coco->pixellib) (4.3.3)\n",
            "Requirement already satisfied: pyrsistent!=0.17.0,!=0.17.1,!=0.17.2,>=0.14.0 in /usr/local/lib/python3.7/dist-packages (from jsonschema>=2.6.0->labelme2coco->pixellib) (0.18.1)\n",
            "Requirement already satisfied: attrs>=17.4.0 in /usr/local/lib/python3.7/dist-packages (from jsonschema>=2.6.0->labelme2coco->pixellib) (21.4.0)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from sahi>=0.8.19->labelme2coco->pixellib) (2.23.0)\n",
            "Requirement already satisfied: fire in /usr/local/lib/python3.7/dist-packages (from sahi>=0.8.19->labelme2coco->pixellib) (0.4.0)\n",
            "Requirement already satisfied: terminaltables in /usr/local/lib/python3.7/dist-packages (from sahi>=0.8.19->labelme2coco->pixellib) (3.1.10)\n",
            "Requirement already satisfied: PyQt5-Qt5>=5.15.2 in /usr/local/lib/python3.7/dist-packages (from pyQt5->pixellib) (5.15.2)\n",
            "Requirement already satisfied: PyQt5-sip<13,>=12.8 in /usr/local/lib/python3.7/dist-packages (from pyQt5->pixellib) (12.10.1)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->sahi>=0.8.19->labelme2coco->pixellib) (2021.10.8)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->sahi>=0.8.19->labelme2coco->pixellib) (3.0.4)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->sahi>=0.8.19->labelme2coco->pixellib) (2.10)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->sahi>=0.8.19->labelme2coco->pixellib) (1.24.3)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!cat /usr/local/lib/python3.7/dist-packages/pixellib/semantic/deeplab.py "
      ],
      "metadata": {
        "id": "GppWozHLRXa6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "files.upload()"
      ],
      "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": 110
        },
        "id": "JniW6_LER4Ty",
        "outputId": "c9e5be8f-6b3e-40be-ee33-06a31bdb2af1"
      },
      "execution_count": 149,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ],
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-7b527ff0-e3a1-4065-805a-214bec19fe54\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-7b527ff0-e3a1-4065-805a-214bec19fe54\">\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 deeplab.py to deeplab.py\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'deeplab.py': b'from __future__ import absolute_import\\nfrom __future__ import division\\nfrom __future__ import print_function\\n\\nimport tensorflow as tf\\n\\nfrom tensorflow.keras.models import Model\\nfrom tensorflow.keras import layers\\nfrom tensorflow.keras.layers import Input\\nfrom tensorflow.keras.layers import Lambda\\nfrom tensorflow.keras.layers import Activation\\nfrom tensorflow.keras.layers import Concatenate\\nfrom tensorflow.keras.layers import Add\\nfrom tensorflow.keras.layers import Dropout\\nfrom tensorflow.keras.layers import BatchNormalization\\nfrom tensorflow.keras.layers import Conv2D\\nfrom tensorflow.keras.layers import DepthwiseConv2D\\nfrom tensorflow.keras.layers import ZeroPadding2D\\nfrom tensorflow.keras.layers import GlobalAveragePooling2D\\nfrom tensorflow.python.keras.utils.layer_utils import get_source_inputs\\nfrom tensorflow.keras import backend as K\\n\\nconfig = tf.compat.v1.ConfigProto()\\nconfig.gpu_options.allow_growth = True\\nsession = tf.compat.v1.InteractiveSession(config=config)\\n\\n\\ndef SepConv_BN(x, filters, prefix, stride=1, kernel_size=3, rate=1, depth_activation=False, epsilon=1e-3):\\n    if stride == 1:\\n        depth_padding = \\'same\\'\\n    else:\\n        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)\\n        pad_total = kernel_size_effective - 1\\n        pad_beg = pad_total // 2\\n        pad_end = pad_total - pad_beg\\n        x = ZeroPadding2D((pad_beg, pad_end))(x)\\n        depth_padding = \\'valid\\'\\n\\n    if not depth_activation:\\n        x = Activation(tf.nn.relu)(x)\\n    x = DepthwiseConv2D((kernel_size, kernel_size), strides=(stride, stride), dilation_rate=(rate, rate),\\n                        padding=depth_padding, use_bias=False, name=prefix + \\'_depthwise\\')(x)\\n    x = BatchNormalization(name=prefix + \\'_depthwise_BN\\', epsilon=epsilon)(x)\\n    if depth_activation:\\n        x = Activation(tf.nn.relu)(x)\\n    x = Conv2D(filters, (1, 1), padding=\\'same\\',\\n               use_bias=False, name=prefix + \\'_pointwise\\')(x)\\n    x = BatchNormalization(name=prefix + \\'_pointwise_BN\\', epsilon=epsilon)(x)\\n    if depth_activation:\\n        x = Activation(tf.nn.relu)(x)\\n\\n    return x\\n\\n\\ndef _conv2d_same(x, filters, prefix, stride=1, kernel_size=3, rate=1):\\n\\n    if stride == 1:\\n        return Conv2D(filters,\\n                      (kernel_size, kernel_size),\\n                      strides=(stride, stride),\\n                      padding=\\'same\\', use_bias=False,\\n                      dilation_rate=(rate, rate),\\n                      name=prefix)(x)\\n    else:\\n        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)\\n        pad_total = kernel_size_effective - 1\\n        pad_beg = pad_total // 2\\n        pad_end = pad_total - pad_beg\\n        x = ZeroPadding2D((pad_beg, pad_end))(x)\\n        return Conv2D(filters,\\n                      (kernel_size, kernel_size),\\n                      strides=(stride, stride),\\n                      padding=\\'valid\\', use_bias=False,\\n                      dilation_rate=(rate, rate),\\n                      name=prefix)(x)\\n\\n\\ndef _xception_block(inputs, depth_list, prefix, skip_connection_type, stride,\\n                    rate=1, depth_activation=False, return_skip=False):\\n    residual = inputs\\n    for i in range(3):\\n        residual = SepConv_BN(residual,\\n                              depth_list[i],\\n                              prefix + \\'_separable_conv{}\\'.format(i + 1),\\n                              stride=stride if i == 2 else 1,\\n                              rate=rate,\\n                              depth_activation=depth_activation)\\n        if i == 1:\\n            skip = residual\\n    if skip_connection_type == \\'conv\\':\\n        shortcut = _conv2d_same(inputs, depth_list[-1], prefix + \\'_shortcut\\',\\n                                kernel_size=1,\\n                                stride=stride)\\n        shortcut = BatchNormalization(name=prefix + \\'_shortcut_BN\\')(shortcut)\\n        outputs = layers.add([residual, shortcut])\\n    elif skip_connection_type == \\'sum\\':\\n        outputs = layers.add([residual, inputs])\\n    elif skip_connection_type == \\'none\\':\\n        outputs = residual\\n    if return_skip:\\n        return outputs, skip\\n    else:\\n        return outputs\\n\\n\\ndef Deeplab_xcep_pascal(weights=None, input_tensor=None, input_shape=(512, 512, 3), num_classes=21, backbone=\\'xception\\',\\n                        OS=8, model_path=\"\", activation=None):\\n\\n    if not (weights in {\\'pascalvoc\\', None}):\\n        raise ValueError(\\'The `weights` argument should be either \\'\\n                         \\'`None` (random initialization), `pascalvoc` \\'\\n                         \\'(pre-trained on PascalVoc)\\')\\n\\n    if input_tensor is None:\\n        img_input = Input(shape=input_shape)\\n    else:\\n        img_input = input_tensor\\n\\n    if OS == 8:\\n\\n        entry_block3_stride = 1\\n        middle_block_rate = 2  # ! Not mentioned in paper, but required\\n        exit_block_rates = (2, 4)\\n        atrous_rates = (12, 24, 36)\\n\\n        x = Conv2D(32, (3, 3), strides=(2, 2),\\n                   name=\\'entry_flow_conv1_1\\', use_bias=False, padding=\\'same\\')(img_input)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_1_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _conv2d_same(x, 64, \\'entry_flow_conv1_2\\', kernel_size=3, stride=1)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_2_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _xception_block(x, [128, 128, 128], \\'entry_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=2,\\n                            depth_activation=False)\\n        x, skip1 = _xception_block(x, [256, 256, 256], \\'entry_flow_block2\\',\\n                                   skip_connection_type=\\'conv\\', stride=2,\\n                                   depth_activation=False, return_skip=True)\\n\\n        x = _xception_block(x, [728, 728, 728], \\'entry_flow_block3\\',\\n                            skip_connection_type=\\'conv\\', stride=entry_block3_stride,\\n                            depth_activation=False)\\n        for i in range(16):\\n            x = _xception_block(x, [728, 728, 728], \\'middle_flow_unit_{}\\'.format(i + 1),\\n                                skip_connection_type=\\'sum\\', stride=1, rate=middle_block_rate,\\n                                depth_activation=False)\\n\\n        x = _xception_block(x, [728, 1024, 1024], \\'exit_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=1, rate=exit_block_rates[0],\\n                            depth_activation=False)\\n        x = _xception_block(x, [1536, 1536, 2048], \\'exit_flow_block2\\',\\n                            skip_connection_type=\\'none\\', stride=1, rate=exit_block_rates[1],\\n                            depth_activation=True)\\n\\n    else:\\n\\n        entry_block3_stride = 2\\n        middle_block_rate = 1\\n        exit_block_rates = (1, 2)\\n        atrous_rates = (6, 12, 18)\\n\\n        x = Conv2D(32, (3, 3), strides=(2, 2),\\n                   name=\\'entry_flow_conv1_1\\', use_bias=False, padding=\\'same\\')(img_input)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_1_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _conv2d_same(x, 64, \\'entry_flow_conv1_2\\', kernel_size=3, stride=1)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_2_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _xception_block(x, [128, 128, 128], \\'entry_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=2,\\n                            depth_activation=False)\\n        x, skip1 = _xception_block(x, [256, 256, 256], \\'entry_flow_block2\\',\\n                                   skip_connection_type=\\'conv\\', stride=2,\\n                                   depth_activation=False, return_skip=True)\\n\\n        x = _xception_block(x, [728, 728, 728], \\'entry_flow_block3\\',\\n                            skip_connection_type=\\'conv\\', stride=entry_block3_stride,\\n                            depth_activation=False)\\n        for i in range(16):\\n            x = _xception_block(x, [728, 728, 728], \\'middle_flow_unit_{}\\'.format(i + 1),\\n                                skip_connection_type=\\'sum\\', stride=1, rate=middle_block_rate,\\n                                depth_activation=False)\\n\\n        x = _xception_block(x, [728, 1024, 1024], \\'exit_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=1, rate=exit_block_rates[0],\\n                            depth_activation=False)\\n        x = _xception_block(x, [1536, 1536, 2048], \\'exit_flow_block2\\',\\n                            skip_connection_type=\\'none\\', stride=1, rate=exit_block_rates[1],\\n                            depth_activation=True)\\n\\n    shape_before = tf.shape(x)\\n    b4 = GlobalAveragePooling2D()(x)\\n    # from (b_size, channels)->(b_size, 1, 1, channels)\\n    b4 = Lambda(lambda x: K.expand_dims(x, 1))(b4)\\n    b4 = Lambda(lambda x: K.expand_dims(x, 1))(b4)\\n    b4 = Conv2D(256, (1, 1), padding=\\'same\\',\\n                use_bias=False, name=\\'image_pooling\\')(b4)\\n    b4 = BatchNormalization(name=\\'image_pooling_BN\\', epsilon=1e-5)(b4)\\n    b4 = Activation(tf.nn.relu)(b4)\\n    # upsample. have to use compat because of the option align_corners\\n    size_before = tf.keras.backend.int_shape(x)\\n    b4 = Lambda(lambda x: tf.compat.v1.image.resize(x, size_before[1:3],\\n                                                    method=\\'bilinear\\', align_corners=True))(b4)\\n    # simple 1x1\\n    b0 = Conv2D(256, (1, 1), padding=\\'same\\', use_bias=False, name=\\'aspp0\\')(x)\\n    b0 = BatchNormalization(name=\\'aspp0_BN\\', epsilon=1e-5)(b0)\\n    b0 = Activation(tf.nn.relu, name=\\'aspp0_activation\\')(b0)\\n\\n    if backbone == \\'xception\\':\\n        # rate = 6 (12)\\n        b1 = SepConv_BN(x, 256, \\'aspp1\\',\\n                        rate=atrous_rates[0], depth_activation=True, epsilon=1e-5)\\n        # rate = 12 (24)\\n        b2 = SepConv_BN(x, 256, \\'aspp2\\',\\n                        rate=atrous_rates[1], depth_activation=True, epsilon=1e-5)\\n        # rate = 18 (36)\\n        b3 = SepConv_BN(x, 256, \\'aspp3\\',\\n                        rate=atrous_rates[2], depth_activation=True, epsilon=1e-5)\\n\\n        # concatenate ASPP branches & project\\n        x = Concatenate()([b4, b0, b1, b2, b3])\\n\\n    x = Conv2D(256, (1, 1), padding=\\'same\\',\\n               use_bias=False, name=\\'concat_projection\\')(x)\\n    x = BatchNormalization(name=\\'concat_projection_BN\\', epsilon=1e-5)(x)\\n    x = Activation(tf.nn.relu)(x)\\n    x = Dropout(0.1)(x)\\n    # DeepLab v.3+ decoder\\n\\n    if backbone == \\'xception\\':\\n        # Feature projection\\n        # x4 (x2) block\\n        size_before2 = tf.keras.backend.int_shape(x)\\n        x = Lambda(lambda xx: tf.compat.v1.image.resize(xx,\\n                                                        skip1.shape[1:3],\\n                                                        method=\\'bilinear\\', align_corners=True))(x)\\n\\n        dec_skip1 = Conv2D(48, (1, 1), padding=\\'same\\',\\n                           use_bias=False, name=\\'feature_projection0\\')(skip1)\\n        dec_skip1 = BatchNormalization(\\n            name=\\'feature_projection0_BN\\', epsilon=1e-5)(dec_skip1)\\n        dec_skip1 = Activation(tf.nn.relu)(dec_skip1)\\n        x = Concatenate()([x, dec_skip1])\\n        x = SepConv_BN(x, 256, \\'decoder_conv0\\',\\n                       depth_activation=True, epsilon=1e-5)\\n        x = SepConv_BN(x, 256, \\'decoder_conv1\\',\\n                       depth_activation=True, epsilon=1e-5)\\n\\n    if (weights == \\'pascalvoc\\' and num_classes == 21):\\n        last_layer_name = \\'logits_semantic\\'\\n    else:\\n        last_layer_name = \\'custom_logits_semantic\\'\\n\\n    x = Conv2D(num_classes, (1, 1), padding=\\'same\\', name=last_layer_name)(x)\\n    size_before3 = tf.keras.backend.int_shape(img_input)\\n    x = Lambda(lambda xx: tf.compat.v1.image.resize(xx,\\n                                                    size_before3[1:3],\\n                                                    method=\\'bilinear\\', align_corners=True))(x)\\n\\n    # Ensure that the model takes into account\\n    # any potential predecessors of `input_tensor`.\\n    if input_tensor is not None:\\n        inputs = get_source_inputs(input_tensor)\\n    else:\\n        inputs = img_input\\n\\n    if activation in {\\'softmax\\', \\'sigmoid\\'}:\\n        x = tf.keras.layers.Activation(activation)(x)\\n\\n    model = Model(inputs, x, name=\\'deeplabv3plus\\')\\n\\n    if weights == \\'pascalvoc\\':\\n        weights_path = model_path\\n        model.load_weights(weights_path, by_name=True)\\n    return model\\n\\n\\ndef Deeplab_xcep_ade20k(weights=None, input_tensor=None, input_shape=(512, 512, 3), num_classes=151, backbone=\\'xception\\',\\n                        OS=8, model_path=\"\", activation=None):\\n\\n    if not (weights in {\\'ade20k\\', None}):\\n        raise ValueError(\\'The `weights` argument should be either \\'\\n                         \\'`None` (random initialization), `ade20k` \\'\\n                         \\'(pre-trained on ADE20K)\\')\\n\\n    if input_tensor is None:\\n        img_input = Input(shape=input_shape)\\n    else:\\n        img_input = input_tensor\\n\\n    if OS == 8:\\n\\n        entry_block3_stride = 1\\n        middle_block_rate = 2  # ! Not mentioned in paper, but required\\n        exit_block_rates = (2, 4)\\n        atrous_rates = (12, 24, 36)\\n\\n        x = Conv2D(32, (3, 3), strides=(2, 2),\\n                   name=\\'entry_flow_conv1_1\\', use_bias=False, padding=\\'same\\')(img_input)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_1_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _conv2d_same(x, 64, \\'entry_flow_conv1_2\\', kernel_size=3, stride=1)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_2_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _xception_block(x, [128, 128, 128], \\'entry_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=2,\\n                            depth_activation=False)\\n        x, skip1 = _xception_block(x, [256, 256, 256], \\'entry_flow_block2\\',\\n                                   skip_connection_type=\\'conv\\', stride=2,\\n                                   depth_activation=False, return_skip=True)\\n\\n        x = _xception_block(x, [728, 728, 728], \\'entry_flow_block3\\',\\n                            skip_connection_type=\\'conv\\', stride=entry_block3_stride,\\n                            depth_activation=False)\\n        for i in range(16):\\n            x = _xception_block(x, [728, 728, 728], \\'middle_flow_unit_{}\\'.format(i + 1),\\n                                skip_connection_type=\\'sum\\', stride=1, rate=middle_block_rate,\\n                                depth_activation=False)\\n\\n        x = _xception_block(x, [728, 1024, 1024], \\'exit_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=1, rate=exit_block_rates[0],\\n                            depth_activation=False)\\n        x = _xception_block(x, [1536, 1536, 2048], \\'exit_flow_block2\\',\\n                            skip_connection_type=\\'none\\', stride=1, rate=exit_block_rates[1],\\n                            depth_activation=True)\\n\\n    else:\\n\\n        entry_block3_stride = 2\\n        middle_block_rate = 1\\n        exit_block_rates = (1, 2)\\n        atrous_rates = (6, 12, 18)\\n\\n        x = Conv2D(32, (3, 3), strides=(2, 2),\\n                   name=\\'entry_flow_conv1_1\\', use_bias=False, padding=\\'same\\')(img_input)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_1_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _conv2d_same(x, 64, \\'entry_flow_conv1_2\\', kernel_size=3, stride=1)\\n        x = BatchNormalization(name=\\'entry_flow_conv1_2_BN\\')(x)\\n        x = Activation(tf.nn.relu)(x)\\n\\n        x = _xception_block(x, [128, 128, 128], \\'entry_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=2,\\n                            depth_activation=False)\\n        x, skip1 = _xception_block(x, [256, 256, 256], \\'entry_flow_block2\\',\\n                                   skip_connection_type=\\'conv\\', stride=2,\\n                                   depth_activation=False, return_skip=True)\\n\\n        x = _xception_block(x, [728, 728, 728], \\'entry_flow_block3\\',\\n                            skip_connection_type=\\'conv\\', stride=entry_block3_stride,\\n                            depth_activation=False)\\n        for i in range(16):\\n            x = _xception_block(x, [728, 728, 728], \\'middle_flow_unit_{}\\'.format(i + 1),\\n                                skip_connection_type=\\'sum\\', stride=1, rate=middle_block_rate,\\n                                depth_activation=False)\\n\\n        x = _xception_block(x, [728, 1024, 1024], \\'exit_flow_block1\\',\\n                            skip_connection_type=\\'conv\\', stride=1, rate=exit_block_rates[0],\\n                            depth_activation=False)\\n        x = _xception_block(x, [1536, 1536, 2048], \\'exit_flow_block2\\',\\n                            skip_connection_type=\\'none\\', stride=1, rate=exit_block_rates[1],\\n                            depth_activation=True)\\n\\n    shape_before = tf.shape(x)\\n    b4 = GlobalAveragePooling2D()(x)\\n    # from (b_size, channels)->(b_size, 1, 1, channels)\\n    b4 = Lambda(lambda x: K.expand_dims(x, 1))(b4)\\n    b4 = Lambda(lambda x: K.expand_dims(x, 1))(b4)\\n    b4 = Conv2D(256, (1, 1), padding=\\'same\\',\\n                use_bias=False, name=\\'image_pooling\\')(b4)\\n    b4 = BatchNormalization(name=\\'image_pooling_BN\\', epsilon=1e-5)(b4)\\n    b4 = Activation(tf.nn.relu)(b4)\\n    # upsample. have to use compat because of the option align_corners\\n    size_before = tf.keras.backend.int_shape(x)\\n    b4 = Lambda(lambda x: tf.compat.v1.image.resize(x, size_before[1:3],\\n                                                    method=\\'bilinear\\', align_corners=True))(b4)\\n    # simple 1x1\\n    b0 = Conv2D(256, (1, 1), padding=\\'same\\', use_bias=False, name=\\'aspp0\\')(x)\\n    b0 = BatchNormalization(name=\\'aspp0_BN\\', epsilon=1e-5)(b0)\\n    b0 = Activation(tf.nn.relu, name=\\'aspp0_activation\\')(b0)\\n\\n    if backbone == \\'xception\\':\\n        # rate = 6 (12)\\n        b1 = SepConv_BN(x, 256, \\'aspp1\\',\\n                        rate=atrous_rates[0], depth_activation=True, epsilon=1e-5)\\n        # rate = 12 (24)\\n        b2 = SepConv_BN(x, 256, \\'aspp2\\',\\n                        rate=atrous_rates[1], depth_activation=True, epsilon=1e-5)\\n        # rate = 18 (36)\\n        b3 = SepConv_BN(x, 256, \\'aspp3\\',\\n                        rate=atrous_rates[2], depth_activation=True, epsilon=1e-5)\\n\\n        # concatenate ASPP branches & project\\n        x = Concatenate()([b4, b0, b1, b2, b3])\\n\\n    x = Conv2D(256, (1, 1), padding=\\'same\\',\\n               use_bias=False, name=\\'concat_projection\\')(x)\\n    x = BatchNormalization(name=\\'concat_projection_BN\\', epsilon=1e-5)(x)\\n    x = Activation(tf.nn.relu)(x)\\n    x = Dropout(0.1)(x)\\n    # DeepLab v.3+ decoder\\n\\n    if backbone == \\'xception\\':\\n        # Feature projection\\n        # x4 (x2) block\\n        size_before2 = tf.keras.backend.int_shape(x)\\n        x = Lambda(lambda xx: tf.compat.v1.image.resize(xx,\\n                                                        skip1.shape[1:3],\\n                                                        method=\\'bilinear\\', align_corners=True))(x)\\n\\n        dec_skip1 = Conv2D(48, (1, 1), padding=\\'same\\',\\n                           use_bias=False, name=\\'feature_projection0\\')(skip1)\\n        dec_skip1 = BatchNormalization(\\n            name=\\'feature_projection0_BN\\', epsilon=1e-5)(dec_skip1)\\n        dec_skip1 = Activation(tf.nn.relu)(dec_skip1)\\n        x = Concatenate()([x, dec_skip1])\\n        x = SepConv_BN(x, 256, \\'decoder_conv0\\',\\n                       depth_activation=True, epsilon=1e-5)\\n        x = SepConv_BN(x, 256, \\'decoder_conv1\\',\\n                       depth_activation=True, epsilon=1e-5)\\n\\n    # if (weights == \\'ade20k\\' and num_classes == 150):\\n    last_layer_name = \\'logits_semantic\\'\\n    # else:\\n    #last_layer_name = \\'custom_logits_semantic\\'\\n\\n    x = Conv2D(num_classes, kernel_size=(1, 1),\\n               padding=\\'same\\', name=last_layer_name)(x)\\n    size_before3 = tf.keras.backend.int_shape(img_input)\\n    x = Lambda(lambda xx: tf.compat.v1.image.resize(xx,\\n                                                    size_before3[1:3],\\n                                                    method=\\'bilinear\\', align_corners=True))(x)\\n\\n    # Ensure that the model takes into account\\n    # any potential predecessors of `input_tensor`.\\n    if input_tensor is not None:\\n        inputs = get_source_inputs(input_tensor)\\n    else:\\n        inputs = img_input\\n\\n    if activation in {\\'softmax\\', \\'sigmoid\\'}:\\n        x = tf.keras.layers.Activation(activation)(x)\\n\\n    model = Model(inputs, x, name=\\'deeplabv3plus\\')\\n\\n    if weights == \\'ade20k\\':\\n        weights_path = model_path\\n        model.load_weights(weights_path, by_name=True)\\n    return model\\n'}"
            ]
          },
          "metadata": {},
          "execution_count": 149
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!mv deeplab.py /usr/local/lib/python3.7/dist-packages/pixellib/semantic/deeplab.py"
      ],
      "metadata": {
        "id": "VPNtZLbiSfEx"
      },
      "execution_count": 150,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 151,
      "metadata": {
        "id": "0wfmp7_dXdcP"
      },
      "outputs": [],
      "source": [
        "from PIL import Image\n",
        "from numpy import asarray\n",
        "from google.colab import files\n",
        "import numpy as np\n",
        "from keras.utils import np_utils\n",
        "import tensorflow as tf\n",
        "import glob\n",
        "from tensorflow.keras import models, layers, optimizers, losses, metrics\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib.image as mpimg\n",
        "import pixellib\n",
        "from pixellib.semantic import semantic_segmentation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VDlDzpPnfDrp",
        "outputId": "06384be8-9225-4bc1-c076-39485d1eff19"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "data  sample_data\n"
          ]
        }
      ],
      "source": [
        "!ls"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "iW2JacC2Yjd2"
      },
      "outputs": [],
      "source": [
        "# X\n",
        "files.upload()"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "mkdir X && mv *.png X"
      ],
      "metadata": {
        "id": "qrcBfTrn5ExE"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lD2-OgbMo35l"
      },
      "outputs": [],
      "source": [
        "# Y\n",
        "files.upload()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "RlDn0l8YsRoG"
      },
      "outputs": [],
      "source": [
        "mkdir Y && mv *.png Y"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Rough Model"
      ],
      "metadata": {
        "id": "Qx5Cn1HFTmwx"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 200,
      "metadata": {
        "id": "YY_ryDebsbW-"
      },
      "outputs": [],
      "source": [
        "x_files = sorted(glob.glob('X/*.png'))\n",
        "y_files = sorted(glob.glob('Y/*.png'))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 201,
      "metadata": {
        "id": "isI8DhPbtGqf"
      },
      "outputs": [],
      "source": [
        "x = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_files])\n",
        "y = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in y_files])"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y = y[:, :, :, np.newaxis]"
      ],
      "metadata": {
        "id": "9HzHRjyNbzzo"
      },
      "execution_count": 202,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 203,
      "metadata": {
        "id": "5BUQt4QvtQAq"
      },
      "outputs": [],
      "source": [
        "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.25, random_state=12)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, x_valid, y_train, y_valid = train_test_split(x_train, y_train, test_size=0.1, random_state=12)"
      ],
      "metadata": {
        "id": "n6KmLo8GTF_Z"
      },
      "execution_count": 204,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "y16G9Dn8h7Uv"
      },
      "outputs": [],
      "source": [
        "rough_model = models.Sequential()\n",
        "\n",
        "# Convolutional layers\n",
        "# model.add(layers.UpSampling2D((2,2)))\n",
        "# model.add(layers.UpSampling2D((2,2)))\n",
        "rough_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "# model.add(layers.MaxPool2D(pool_size=(2, 2)))\n",
        "rough_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "rough_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "rough_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "# model.add(layers.MaxPool2D(pool_size=(2, 2)))\n",
        "rough_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "# model.add(layers.MaxPool2D(pool_size=(2, 2)))\n",
        "rough_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "# model.add(layers.MaxPool2D(pool_size=(2, 2)))\n",
        "rough_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "# model.add(layers.MaxPool2D(pool_size=(2, 2)))\n",
        "rough_model.add(layers.Conv2DTranspose(filters=50, padding=\"same\",  kernel_size=(3, 2)))\n",
        "rough_model.add(layers.BatchNormalization())\n",
        "rough_model.add(layers.Conv2D(filters=1, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "rough_model.compile(loss=losses.MeanSquaredError(),\n",
        "                  optimizer= optimizers.Adam(learning_rate=0.01),\n",
        "                  metrics=[metrics.MeanSquaredError()]) "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VanDTubTnGgg",
        "outputId": "472f0b07-88cd-430b-f945-d37bc893eaeb"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "68/68 [==============================] - 90s 1s/step - loss: 1865978.3750 - mean_squared_error: 1865978.8750 - val_loss: 1539729536.0000 - val_mean_squared_error: 1539729536.0000\n",
            "Epoch 2/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 696310.5000 - mean_squared_error: 696310.7500 - val_loss: 15902929.0000 - val_mean_squared_error: 15902929.0000\n",
            "Epoch 3/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 329871.0625 - mean_squared_error: 329871.0625 - val_loss: 24398938.0000 - val_mean_squared_error: 24398940.0000\n",
            "Epoch 4/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 266747.7188 - mean_squared_error: 266747.7500 - val_loss: 12623595.0000 - val_mean_squared_error: 12623593.0000\n",
            "Epoch 5/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 257320.2031 - mean_squared_error: 257320.1406 - val_loss: 5329496.5000 - val_mean_squared_error: 5329496.5000\n",
            "Epoch 6/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 255685.2969 - mean_squared_error: 255685.3750 - val_loss: 1832021.7500 - val_mean_squared_error: 1832021.8750\n",
            "Epoch 7/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 253943.4688 - mean_squared_error: 253943.4531 - val_loss: 2093328.5000 - val_mean_squared_error: 2093328.5000\n",
            "Epoch 8/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 259147.5312 - mean_squared_error: 259147.5312 - val_loss: 1616237.8750 - val_mean_squared_error: 1616237.8750\n",
            "Epoch 9/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 256306.6406 - mean_squared_error: 256306.6406 - val_loss: 568146.1250 - val_mean_squared_error: 568146.1250\n",
            "Epoch 10/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 253036.7656 - mean_squared_error: 253036.8594 - val_loss: 1724097.2500 - val_mean_squared_error: 1724097.0000\n",
            "Epoch 11/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 250698.7188 - mean_squared_error: 250698.7344 - val_loss: 668036.2500 - val_mean_squared_error: 668036.1875\n",
            "Epoch 12/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 246406.4688 - mean_squared_error: 246406.3906 - val_loss: 420249.9688 - val_mean_squared_error: 420249.9688\n",
            "Epoch 13/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 246939.2969 - mean_squared_error: 246939.2812 - val_loss: 1704955.3750 - val_mean_squared_error: 1704955.5000\n",
            "Epoch 14/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 251294.9062 - mean_squared_error: 251294.8438 - val_loss: 350068.0625 - val_mean_squared_error: 350068.0938\n",
            "Epoch 15/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 248931.2031 - mean_squared_error: 248931.1875 - val_loss: 400044.4062 - val_mean_squared_error: 400044.4062\n",
            "Epoch 16/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 246637.9062 - mean_squared_error: 246637.9531 - val_loss: 347312.7188 - val_mean_squared_error: 347312.7188\n",
            "Epoch 17/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 243029.8594 - mean_squared_error: 243029.8438 - val_loss: 345873.4375 - val_mean_squared_error: 345873.4688\n",
            "Epoch 18/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 240709.2344 - mean_squared_error: 240709.2500 - val_loss: 350682.1562 - val_mean_squared_error: 350682.1875\n",
            "Epoch 19/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 237011.7500 - mean_squared_error: 237011.7500 - val_loss: 574118.5625 - val_mean_squared_error: 574118.5000\n",
            "Epoch 20/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 241408.4219 - mean_squared_error: 241408.4375 - val_loss: 530755.9375 - val_mean_squared_error: 530755.9375\n",
            "Epoch 21/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 237630.1562 - mean_squared_error: 237630.1250 - val_loss: 404068.9375 - val_mean_squared_error: 404068.9375\n",
            "Epoch 22/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 236796.1562 - mean_squared_error: 236796.1250 - val_loss: 761795.9375 - val_mean_squared_error: 761796.0000\n",
            "Epoch 23/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 222544.7344 - mean_squared_error: 222544.7500 - val_loss: 430553.7812 - val_mean_squared_error: 430553.7812\n",
            "Epoch 24/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 227069.0000 - mean_squared_error: 227068.9844 - val_loss: 475176.5000 - val_mean_squared_error: 475176.5000\n",
            "Epoch 25/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 223923.1406 - mean_squared_error: 223923.1875 - val_loss: 554459.8125 - val_mean_squared_error: 554459.8750\n",
            "Epoch 26/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 217177.9531 - mean_squared_error: 217177.9375 - val_loss: 894452.1875 - val_mean_squared_error: 894452.1875\n",
            "Epoch 27/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 218859.5781 - mean_squared_error: 218859.5781 - val_loss: 863090.2500 - val_mean_squared_error: 863090.3125\n",
            "Epoch 28/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 216740.5938 - mean_squared_error: 216740.6250 - val_loss: 547849.1875 - val_mean_squared_error: 547849.1875\n",
            "Epoch 29/30\n",
            "68/68 [==============================] - 74s 1s/step - loss: 217322.8125 - mean_squared_error: 217322.7969 - val_loss: 435823.4375 - val_mean_squared_error: 435823.4062\n",
            "Epoch 30/30\n",
            "68/68 [==============================] - 73s 1s/step - loss: 231514.2812 - mean_squared_error: 231514.2344 - val_loss: 362978.4062 - val_mean_squared_error: 362978.4062\n"
          ]
        }
      ],
      "source": [
        "history = rough_model.fit(x_train, y_train, epochs=30, batch_size=2,\n",
        "                  validation_data=(x_valid, y_valid))"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "rough_model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BgCTjM2C30yZ",
        "outputId": "35b0ab5a-1e35-4e6a-f65d-b660844234d1"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d_16 (Conv2D)          (None, 960, 720, 50)      1400      \n",
            "                                                                 \n",
            " batch_normalization_16 (Bat  (None, 960, 720, 50)     200       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_17 (Conv2D)          (None, 960, 720, 50)      22550     \n",
            "                                                                 \n",
            " batch_normalization_17 (Bat  (None, 960, 720, 50)     200       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_18 (Conv2D)          (None, 960, 720, 80)      36080     \n",
            "                                                                 \n",
            " batch_normalization_18 (Bat  (None, 960, 720, 80)     320       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_19 (Conv2D)          (None, 960, 720, 120)     86520     \n",
            "                                                                 \n",
            " batch_normalization_19 (Bat  (None, 960, 720, 120)    480       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_20 (Conv2D)          (None, 960, 720, 120)     129720    \n",
            "                                                                 \n",
            " batch_normalization_20 (Bat  (None, 960, 720, 120)    480       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_21 (Conv2D)          (None, 960, 720, 80)      86480     \n",
            "                                                                 \n",
            " batch_normalization_21 (Bat  (None, 960, 720, 80)     320       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_22 (Conv2D)          (None, 960, 720, 50)      36050     \n",
            "                                                                 \n",
            " batch_normalization_22 (Bat  (None, 960, 720, 50)     200       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_transpose_2 (Conv2DT  (None, 960, 720, 50)     15050     \n",
            " ranspose)                                                       \n",
            "                                                                 \n",
            " batch_normalization_23 (Bat  (None, 960, 720, 50)     200       \n",
            " chNormalization)                                                \n",
            "                                                                 \n",
            " conv2d_23 (Conv2D)          (None, 960, 720, 1)       451       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 416,701\n",
            "Trainable params: 415,501\n",
            "Non-trainable params: 1,200\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 293
        },
        "id": "hf6QHQ9j4PPa",
        "outputId": "9be25f1b-5a16-40c6-a07b-d3a42a264c72"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f05f03dfb90>]"
            ]
          },
          "metadata": {},
          "execution_count": 19
        },
        {
          "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.plot(history.history['val_loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 293
        },
        "id": "ty1joLNQl0FC",
        "outputId": "b23a2ea2-1c0d-4a43-e612-bd713ec7e8ff"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f068a5b1550>]"
            ]
          },
          "metadata": {},
          "execution_count": 20
        },
        {
          "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": [
        "evaluation = rough_model.evaluate(x_test[:10], y_test[:10])"
      ],
      "metadata": {
        "id": "HRf6-e2nr36l",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bcfcf210-9417-454e-e695-3ef4bbd85424"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 55s 55s/step - loss: 299625.8438 - mean_squared_error: 299625.8438\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.gray()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "ExuyjV2Q5vHi",
        "outputId": "453a8054-cce6-4e1c-bcb7-567022cd2624"
      },
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 0 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(x[15:16][0].astype(np.int32), interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "xMB4qgoa6GuP",
        "outputId": "9956dfe1-9922-4f13-feb9-1b023e7e061a"
      },
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f04f6af7d10>"
            ]
          },
          "metadata": {},
          "execution_count": 46
        },
        {
          "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(y[15:16][0].astype(np.int32), interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "KfI2NPmg7Azg",
        "outputId": "5b6169d4-e655-4215-8809-07682b3fab8f"
      },
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f04f6ad7c10>"
            ]
          },
          "metadata": {},
          "execution_count": 47
        },
        {
          "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": [
        "prediction = rough_model.predict(x[15:16])\n",
        "plt.imshow(np.squeeze(prediction[0]).astype(np.int32) ,interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "rE0o63SX5fiq",
        "outputId": "150b5067-8237-45a3-c4d0-146f263c1efe"
      },
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f04f6a5abd0>"
            ]
          },
          "metadata": {},
          "execution_count": 48
        },
        {
          "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": [
        "for k in range(len(x)):\n",
        "  I = rough_model.predict(x[k: k + 1])[0]\n",
        "  I = np.squeeze(I)\n",
        "  I8 = (((I - I.min()) / (I.max() - I.min())) * 255.9).astype(np.uint8)\n",
        "  img = Image.fromarray(I8)\n",
        "  img.save(x_files[k][2:])"
      ],
      "metadata": {
        "id": "ykrygnVk4PHf"
      },
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "mkdir rough_X & mv *.png rough_X"
      ],
      "metadata": {
        "id": "fv2O0gsO9Jg1"
      },
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "files.upload()"
      ],
      "metadata": {
        "id": "dnf4MGzU9_fj"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "mkdir Y_edges & mv *.png Y_edges"
      ],
      "metadata": {
        "id": "nECoedEt_FeR"
      },
      "execution_count": 62,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Gradient Model"
      ],
      "metadata": {
        "id": "LYqA6IA0TrvS"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough_files = sorted(glob.glob('rough_X/*.png'))\n",
        "x_color_files = sorted(glob.glob('X/*.png'))\n",
        "y_edges_files = sorted(glob.glob('Y_edges/*.png'))"
      ],
      "metadata": {
        "id": "-uujeX68_KmJ"
      },
      "execution_count": 63,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_rough_files])\n",
        "x_color = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_color_files])\n",
        "y_edge = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in y_edges_files])"
      ],
      "metadata": {
        "id": "aYERS6Cy_bLj"
      },
      "execution_count": 72,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough = x_rough[:, :, :, np.newaxis]\n",
        "y_edge = y_edge[:, :, :, np.newaxis]"
      ],
      "metadata": {
        "id": "cR9CGPHaACbh"
      },
      "execution_count": 75,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x = np.array(tf.keras.layers.Concatenate()([x_rough, x_color]))"
      ],
      "metadata": {
        "id": "H1YkvHAm_5KB"
      },
      "execution_count": 82,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "y = y_edge"
      ],
      "metadata": {
        "id": "iC1VzplsA2DY"
      },
      "execution_count": 85,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.25, random_state=12)"
      ],
      "metadata": {
        "id": "omqSNLMS_2xK"
      },
      "execution_count": 86,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, x_valid, y_train, y_valid = train_test_split(x_train, y_train, test_size=0.1, random_state=12)"
      ],
      "metadata": {
        "id": "bsG7bEGgAyrA"
      },
      "execution_count": 87,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "edge_model = models.Sequential()\n",
        "\n",
        "# Convolutional layers\n",
        "edge_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2DTranspose(filters=50, padding=\"same\",  kernel_size=(3, 2)))\n",
        "edge_model.add(layers.BatchNormalization())\n",
        "edge_model.add(layers.Conv2D(filters=1, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "edge_model.compile(loss=losses.MeanSquaredError(),\n",
        "                  optimizer= optimizers.Adam(learning_rate=0.01),\n",
        "                  metrics=[metrics.MeanSquaredError()]) "
      ],
      "metadata": {
        "id": "6GZnI0T2A7Sk"
      },
      "execution_count": 102,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "history = edge_model.fit(x_train, y_train, epochs=30, batch_size=2,\n",
        "                  validation_data=(x_valid, y_valid))"
      ],
      "metadata": {
        "id": "6SFUSgDkDaKk"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "hyQNINtrL-jv",
        "outputId": "ff2ebc9b-ec16-4a1d-d519-1773b354c35a"
      },
      "execution_count": 104,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f068a6b7250>]"
            ]
          },
          "metadata": {},
          "execution_count": 104
        },
        {
          "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.plot(history.history['val_loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "HR6sH8AFMF-J",
        "outputId": "0d33323b-72db-4200-de92-44ed7bd98eb4"
      },
      "execution_count": 105,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f068a6817d0>]"
            ]
          },
          "metadata": {},
          "execution_count": 105
        },
        {
          "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": [
        "evaluation = edge_model.evaluate(x_test[:5], y_test[:5])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BXx5g2rZMKeJ",
        "outputId": "2e42b333-1b90-40ce-c178-c5e4ffcee67a"
      },
      "execution_count": 107,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 11s 11s/step - loss: 6.7253 - mean_squared_error: 6.7253\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(x[15:16][0].astype(np.int32), interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "77Ds86DdMt_W",
        "outputId": "e78311de-916c-4d61-d09a-f8734d6deb38"
      },
      "execution_count": 108,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f068a7bd050>"
            ]
          },
          "metadata": {},
          "execution_count": 108
        },
        {
          "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(np.squeeze(y[15:16][0].astype(np.int32)), interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "2BjSGwe_Mwcm",
        "outputId": "da334bbd-fc9c-429a-d417-47ddd8413a9c"
      },
      "execution_count": 110,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f06a0191610>"
            ]
          },
          "metadata": {},
          "execution_count": 110
        },
        {
          "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": [
        "prediction = edge_model.predict(x[15:16])\n",
        "plt.imshow(np.squeeze(prediction[0]).astype(np.int32) ,interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "69z4zOrdM4dp",
        "outputId": "b0f27007-62f1-4e1f-a46c-67086f06fdce"
      },
      "execution_count": 112,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f04f6596b50>"
            ]
          },
          "metadata": {},
          "execution_count": 112
        },
        {
          "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": [
        "for k in range(len(x)):\n",
        "  I = edge_model.predict(x[k: k + 1])[0]\n",
        "  I = np.squeeze(I)\n",
        "  I8 = (((I - I.min()) / (I.max() - I.min())) * 255.9).astype(np.uint32)\n",
        "  img = Image.fromarray(I8)\n",
        "  img.save(x_files[k][2:])"
      ],
      "metadata": {
        "id": "_x4-IgYcNIBe"
      },
      "execution_count": 113,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!mkdir Y_edge_detection & mv *.png Y_edge_detection"
      ],
      "metadata": {
        "id": "-FLQdD-lNzev"
      },
      "execution_count": 116,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!mkdir image_segment"
      ],
      "metadata": {
        "id": "xj1_59-KOA_F"
      },
      "execution_count": 118,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Segment Model"
      ],
      "metadata": {
        "id": "5Zx2rz-ZTxk8"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!wget https://github.com/ayoolaolafenwa/PixelLib/releases/download/1.1/deeplabv3_xception_tf_dim_ordering_tf_kernels.h5"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IryWSGqQOUK7",
        "outputId": "33f5c78c-2b00-4998-e17f-d46845bd309c"
      },
      "execution_count": 136,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2022-05-04 18:05:32--  https://github.com/ayoolaolafenwa/PixelLib/releases/download/1.1/deeplabv3_xception_tf_dim_ordering_tf_kernels.h5\n",
            "Resolving github.com (github.com)... 20.205.243.166\n",
            "Connecting to github.com (github.com)|20.205.243.166|:443... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://objects.githubusercontent.com/github-production-release-asset-2e65be/255074156/53c11380-90ee-11ea-905d-412859743640?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20220504%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20220504T180532Z&X-Amz-Expires=300&X-Amz-Signature=7c7457e77e4a9a377990d244c1286d791556341fb6b81bae650e3be0a41c7ed2&X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=255074156&response-content-disposition=attachment%3B%20filename%3Ddeeplabv3_xception_tf_dim_ordering_tf_kernels.h5&response-content-type=application%2Foctet-stream [following]\n",
            "--2022-05-04 18:05:33--  https://objects.githubusercontent.com/github-production-release-asset-2e65be/255074156/53c11380-90ee-11ea-905d-412859743640?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20220504%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20220504T180532Z&X-Amz-Expires=300&X-Amz-Signature=7c7457e77e4a9a377990d244c1286d791556341fb6b81bae650e3be0a41c7ed2&X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=255074156&response-content-disposition=attachment%3B%20filename%3Ddeeplabv3_xception_tf_dim_ordering_tf_kernels.h5&response-content-type=application%2Foctet-stream\n",
            "Resolving objects.githubusercontent.com (objects.githubusercontent.com)... 185.199.108.133, 185.199.109.133, 185.199.110.133, ...\n",
            "Connecting to objects.githubusercontent.com (objects.githubusercontent.com)|185.199.108.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 166096424 (158M) [application/octet-stream]\n",
            "Saving to: ‘deeplabv3_xception_tf_dim_ordering_tf_kernels.h5’\n",
            "\n",
            "deeplabv3_xception_ 100%[===================>] 158.40M  4.80MB/s    in 28s     \n",
            "\n",
            "2022-05-04 18:06:01 (5.65 MB/s) - ‘deeplabv3_xception_tf_dim_ordering_tf_kernels.h5’ saved [166096424/166096424]\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!ls"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QMQRM2erOEKd",
        "outputId": "1d5099c4-24c4-488a-eaf7-a86d3f69e287"
      },
      "execution_count": 128,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "data\t       mask_rcnn_coco.h5  sample_data  Y\t\t Y_edges\n",
            "image_segment  rough_X\t\t  X\t       Y_edge_detection\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "segment_image = semantic_segmentation()\n",
        "segment_image.load_pascalvoc_model(\"deeplabv3_xception_tf_dim_ordering_tf_kernels.h5\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zdocL7YUN-x8",
        "outputId": "ae902f27-6f71-4ecf-c740-321f9e1d4b08"
      },
      "execution_count": 152,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.7/dist-packages/keras/layers/normalization/batch_normalization.py:532: _colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x_files = sorted(glob.glob('X/*.png'))"
      ],
      "metadata": {
        "id": "j4UCmTnRTMw-"
      },
      "execution_count": 155,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "for k in range(len(x_files)):\n",
        "  segment_image.segmentAsPascalvoc(x_files[k], output_image_name = \"image_segment/\" + x_files[k][2:])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yJgDyxp9PMyQ",
        "outputId": "ea2e2023-fe7d-4092-c1ba-b433f357aac5"
      },
      "execution_count": 158,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n",
            "Processed Image saved successfuly in your current working directory.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Fine Model"
      ],
      "metadata": {
        "id": "MbvglhXOT23A"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!ls"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vIFc8MIDT_Rh",
        "outputId": "7028f196-7db4-49d7-d986-176d99566400"
      },
      "execution_count": 159,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "data\t\t\t\t\t\t  rough_X      Y_edge_detection\n",
            "deeplabv3_xception_tf_dim_ordering_tf_kernels.h5  sample_data  Y_edges\n",
            "image_segment\t\t\t\t\t  X\n",
            "mask_rcnn_coco.h5\t\t\t\t  Y\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough_files = sorted(glob.glob('rough_X/*.png'))\n",
        "x_color_files = sorted(glob.glob('X/*.png'))\n",
        "x_edges_detect_files = sorted(glob.glob('Y_edge_detection/*.png'))\n",
        "x_image_segment_files = sorted(glob.glob('image_segment/*.png'))\n",
        "y_files = sorted(glob.glob('Y/*.png'))"
      ],
      "metadata": {
        "id": "poNBTpMf4OEV"
      },
      "execution_count": 160,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_rough_files])\n",
        "x_color = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_color_files])\n",
        "x_edges_detect = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_edges_detect_files])\n",
        "x_image_segment = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in x_image_segment_files])\n",
        "y = np.array([np.array(Image.open(fname)).astype(np.float16) for fname in y_files])"
      ],
      "metadata": {
        "id": "TleYUmdGUPUk"
      },
      "execution_count": 161,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_rough = x_rough[:, :, :, np.newaxis]\n",
        "x_edges_detect = x_edges_detect[:, :, :, np.newaxis]\n",
        "y = y[:, :, :, np.newaxis]"
      ],
      "metadata": {
        "id": "R6piX2siUjeB"
      },
      "execution_count": 167,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x = np.concatenate((x_rough, x_color, x_edges_detect, x_image_segment), axis=3)"
      ],
      "metadata": {
        "id": "w-BYwHJHUsex"
      },
      "execution_count": 180,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.25, random_state=12)"
      ],
      "metadata": {
        "id": "_pia9sOLVNkJ"
      },
      "execution_count": 182,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train, x_valid, y_train, y_valid = train_test_split(x_train, y_train, test_size=0.1, random_state=12)"
      ],
      "metadata": {
        "id": "YLwY75ArV08v"
      },
      "execution_count": 183,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "fine_tune_model = models.Sequential()\n",
        "\n",
        "# Convolutional layers\n",
        "fine_tune_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=120, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=80, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=50, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2DTranspose(filters=50, padding=\"same\",  kernel_size=(3, 2)))\n",
        "fine_tune_model.add(layers.BatchNormalization())\n",
        "fine_tune_model.add(layers.Conv2D(filters=1, padding=\"same\", kernel_size=(3,3), activation='relu'))\n",
        "fine_tune_model.compile(loss=losses.MeanSquaredError(),\n",
        "                  optimizer= optimizers.Adam(learning_rate=0.01),\n",
        "                  metrics=[metrics.MeanSquaredError()]) "
      ],
      "metadata": {
        "id": "8YBYPsu3V1OJ"
      },
      "execution_count": 184,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "history = fine_tune_model.fit(x_train, y_train, epochs=30, batch_size=2,\n",
        "                  validation_data=(x_valid, y_valid))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WLX8tnxWV9Rl",
        "outputId": "f37e7c99-9355-447c-f39e-fefeec01c906"
      },
      "execution_count": 185,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Train on 135 samples, validate on 15 samples\n",
            "Epoch 1/30\n",
            "135/135 [==============================] - ETA: 0s - loss: 1873940.4361 - mean_squared_error: 1873940.1250"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/engine/training_v1.py:2057: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
            "  updates = self.state_updates\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r135/135 [==============================] - 108s 796ms/sample - loss: 1873940.4361 - mean_squared_error: 1873940.1250 - val_loss: 27369754.8000 - val_mean_squared_error: 27369752.0000\n",
            "Epoch 2/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 675012.1934 - mean_squared_error: 675012.1875 - val_loss: 2574587.4625 - val_mean_squared_error: 2574587.7500\n",
            "Epoch 3/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 283966.8257 - mean_squared_error: 283966.8750 - val_loss: 10461957.2667 - val_mean_squared_error: 10461957.0000\n",
            "Epoch 4/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 236228.1303 - mean_squared_error: 236228.1250 - val_loss: 4557709.4000 - val_mean_squared_error: 4557710.0000\n",
            "Epoch 5/30\n",
            "135/135 [==============================] - 105s 774ms/sample - loss: 233016.8100 - mean_squared_error: 233016.8438 - val_loss: 689942.0167 - val_mean_squared_error: 689942.0000\n",
            "Epoch 6/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 221359.3500 - mean_squared_error: 221359.3281 - val_loss: 2021326.6583 - val_mean_squared_error: 2021326.6250\n",
            "Epoch 7/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 227872.3436 - mean_squared_error: 227872.3281 - val_loss: 1542076.5500 - val_mean_squared_error: 1542076.5000\n",
            "Epoch 8/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 227275.9639 - mean_squared_error: 227276.0000 - val_loss: 534251.1458 - val_mean_squared_error: 534251.1250\n",
            "Epoch 9/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 230529.8785 - mean_squared_error: 230529.8906 - val_loss: 363221.9865 - val_mean_squared_error: 363222.0000\n",
            "Epoch 10/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 221628.2822 - mean_squared_error: 221628.2812 - val_loss: 358176.5917 - val_mean_squared_error: 358176.6250\n",
            "Epoch 11/30\n",
            "135/135 [==============================] - 105s 775ms/sample - loss: 212688.8425 - mean_squared_error: 212688.8438 - val_loss: 476153.7583 - val_mean_squared_error: 476153.7812\n",
            "Epoch 12/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 217971.0163 - mean_squared_error: 217970.9844 - val_loss: 465871.2000 - val_mean_squared_error: 465871.1875\n",
            "Epoch 13/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 215049.0503 - mean_squared_error: 215049.0312 - val_loss: 415494.9068 - val_mean_squared_error: 415494.9062\n",
            "Epoch 14/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 215511.7823 - mean_squared_error: 215511.8125 - val_loss: 289639.5573 - val_mean_squared_error: 289639.5312\n",
            "Epoch 15/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 212283.8606 - mean_squared_error: 212283.9219 - val_loss: 323487.6620 - val_mean_squared_error: 323487.6562\n",
            "Epoch 16/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 196276.3581 - mean_squared_error: 196276.3438 - val_loss: 542557.8385 - val_mean_squared_error: 542557.8750\n",
            "Epoch 17/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 198820.6907 - mean_squared_error: 198820.6719 - val_loss: 459620.4667 - val_mean_squared_error: 459620.4688\n",
            "Epoch 18/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 181892.9675 - mean_squared_error: 181892.9844 - val_loss: 414725.8161 - val_mean_squared_error: 414725.8438\n",
            "Epoch 19/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 210392.0812 - mean_squared_error: 210392.0625 - val_loss: 443717.3732 - val_mean_squared_error: 443717.3438\n",
            "Epoch 20/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 187209.9314 - mean_squared_error: 187209.9062 - val_loss: 469310.3479 - val_mean_squared_error: 469310.3750\n",
            "Epoch 21/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 184908.5071 - mean_squared_error: 184908.5000 - val_loss: 639682.7336 - val_mean_squared_error: 639682.7500\n",
            "Epoch 22/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 194549.7135 - mean_squared_error: 194549.7500 - val_loss: 906455.8667 - val_mean_squared_error: 906455.8125\n",
            "Epoch 23/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 189080.3520 - mean_squared_error: 189080.3438 - val_loss: 1243312.2708 - val_mean_squared_error: 1243312.2500\n",
            "Epoch 24/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 186699.5819 - mean_squared_error: 186699.5625 - val_loss: 731413.0938 - val_mean_squared_error: 731413.0625\n",
            "Epoch 25/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 189810.5660 - mean_squared_error: 189810.5469 - val_loss: 406823.3333 - val_mean_squared_error: 406823.3438\n",
            "Epoch 26/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 168888.7304 - mean_squared_error: 168888.7500 - val_loss: 282711.8271 - val_mean_squared_error: 282711.8438\n",
            "Epoch 27/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 176096.9760 - mean_squared_error: 176097.0000 - val_loss: 357030.1750 - val_mean_squared_error: 357030.1875\n",
            "Epoch 28/30\n",
            "135/135 [==============================] - 105s 777ms/sample - loss: 170338.0400 - mean_squared_error: 170338.0156 - val_loss: 555222.0557 - val_mean_squared_error: 555222.1250\n",
            "Epoch 29/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 182075.1399 - mean_squared_error: 182075.0938 - val_loss: 850546.9198 - val_mean_squared_error: 850546.9375\n",
            "Epoch 30/30\n",
            "135/135 [==============================] - 105s 776ms/sample - loss: 167849.1167 - mean_squared_error: 167849.0781 - val_loss: 251770.9052 - val_mean_squared_error: 251770.9062\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 293
        },
        "id": "54GxoFMDWBUu",
        "outputId": "ad869f8a-202d-4d5b-fc45-34ca3577d76c"
      },
      "execution_count": 186,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7efeb86a9990>]"
            ]
          },
          "metadata": {},
          "execution_count": 186
        },
        {
          "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.plot(history.history['val_loss'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 293
        },
        "id": "mQhoOg0NiH3Y",
        "outputId": "4327f586-1fd7-411b-9b16-658ff5c85844"
      },
      "execution_count": 187,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7efeb8616590>]"
            ]
          },
          "metadata": {},
          "execution_count": 187
        },
        {
          "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": [
        "prediction = fine_tune_model.predict(x[15:16])\n",
        "plt.imshow(np.squeeze(prediction[0]).astype(np.int32) ,interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "a3tx8N1IiJc3",
        "outputId": "64e54f51-f80e-4ef6-acf5-b2b35b566dff"
      },
      "execution_count": 190,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7efeb8586910>"
            ]
          },
          "metadata": {},
          "execution_count": 190
        },
        {
          "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(np.squeeze(y[15:16][0].astype(np.int32)), interpolation='nearest')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "V9sQ5sceifJD",
        "outputId": "d99494ca-7f3d-41e4-a59e-80d0021f776c"
      },
      "execution_count": 192,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7efeb849da90>"
            ]
          },
          "metadata": {},
          "execution_count": 192
        },
        {
          "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": [
        "fine_tune_evaluation = fine_tune_model.evaluate(x_test[:5], y_test[:5])"
      ],
      "metadata": {
        "id": "0p3yticPkgBG",
        "outputId": "98803263-db93-4d2a-91dc-85767d7171e6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 208,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[160350.359375, 160350.36]"
            ]
          },
          "metadata": {},
          "execution_count": 208
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fine_tune_evaluation"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UvQsIIaHjhcj",
        "outputId": "a65f232d-e2a1-4789-83b9-cd4e5bb7bec7"
      },
      "execution_count": 209,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[160350.359375, 160350.36]"
            ]
          },
          "metadata": {},
          "execution_count": 209
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "t3pDG8CtkiSL"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "machine_shape": "hm",
      "name": "Depth Data 2D Keras.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}