TF2/tf2_tiff_images_dataset_pipeline.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "tf2-tiff-images-dataset-pipeline",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [
        "UnOhpJ-PR6HN"
      ],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/lc0/deeplearning-playground/blob/master/TF2/tf2_tiff_images_dataset_pipeline.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bahskCs761an",
        "colab_type": "code",
        "outputId": "c82e2815-7d66-4e7c-a26f-fe5df059785f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 632
        }
      },
      "source": [
        "!pip install tf-nightly-gpu-2.0-preview"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting tf-nightly-gpu-2.0-preview\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/75/90/a45df7a91472cc128beb0a9e4824aeb1c375bce2f35fd034fcf609118364/tf_nightly_gpu_2.0_preview-2.0.0.dev20190628-cp36-cp36m-manylinux1_x86_64.whl (369.5MB)\n",
            "\u001b[K     |████████████████████████████████| 369.5MB 51kB/s \n",
            "\u001b[?25hCollecting tensorflow-estimator-2.0-preview (from tf-nightly-gpu-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/15/2a/48aa97225aa45b9a53ddc26164952b32fab5d3f41e275617877f71f70163/tensorflow_estimator_2.0_preview-1.14.0.dev2019062700-py2.py3-none-any.whl (447kB)\n",
            "\u001b[K     |████████████████████████████████| 450kB 37.1MB/s \n",
            "\u001b[?25hRequirement already satisfied: six>=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.12.0)\n",
            "Requirement already satisfied: absl-py>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (0.7.1)\n",
            "Requirement already satisfied: google-pasta>=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (0.1.7)\n",
            "Requirement already satisfied: protobuf>=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (3.7.1)\n",
            "Requirement already satisfied: termcolor>=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.1.0)\n",
            "Requirement already satisfied: wrapt>=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.11.1)\n",
            "Requirement already satisfied: keras-applications>=1.0.8 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.0.8)\n",
            "Requirement already satisfied: keras-preprocessing>=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.1.0)\n",
            "Requirement already satisfied: astor>=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (0.8.0)\n",
            "Requirement already satisfied: grpcio>=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.15.0)\n",
            "Requirement already satisfied: wheel>=0.26 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (0.33.4)\n",
            "Collecting opt-einsum>=2.3.2 (from tf-nightly-gpu-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/f6/d6/44792ec668bcda7d91913c75237314e688f70415ab2acd7172c845f0b24f/opt_einsum-2.3.2.tar.gz (59kB)\n",
            "\u001b[K     |████████████████████████████████| 61kB 19.0MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy<2.0,>=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (1.16.4)\n",
            "Collecting tb-nightly<1.15.0a0,>=1.14.0a0 (from tf-nightly-gpu-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/3a/60/afa129c3621d62c885599076f8e89737d66e5dfffad1a08842b1c11b4540/tb_nightly-1.14.0a20190614-py3-none-any.whl (3.1MB)\n",
            "\u001b[K     |████████████████████████████████| 3.2MB 29.8MB/s \n",
            "\u001b[?25hRequirement already satisfied: gast>=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tf-nightly-gpu-2.0-preview) (0.2.2)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.6/dist-packages (from protobuf>=3.6.1->tf-nightly-gpu-2.0-preview) (41.0.1)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications>=1.0.8->tf-nightly-gpu-2.0-preview) (2.8.0)\n",
            "Requirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tb-nightly<1.15.0a0,>=1.14.0a0->tf-nightly-gpu-2.0-preview) (3.1.1)\n",
            "Requirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tb-nightly<1.15.0a0,>=1.14.0a0->tf-nightly-gpu-2.0-preview) (0.15.4)\n",
            "Building wheels for collected packages: opt-einsum\n",
            "  Building wheel for opt-einsum (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Stored in directory: /root/.cache/pip/wheels/51/3e/a3/b351fae0cbf15373c2136a54a70f43fea5fe91d8168a5faaa4\n",
            "Successfully built opt-einsum\n",
            "Installing collected packages: tensorflow-estimator-2.0-preview, opt-einsum, tb-nightly, tf-nightly-gpu-2.0-preview\n",
            "Successfully installed opt-einsum-2.3.2 tb-nightly-1.14.0a20190614 tensorflow-estimator-2.0-preview-1.14.0.dev2019062700 tf-nightly-gpu-2.0-preview-2.0.0.dev20190628\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-evSwzeFR0_K",
        "colab_type": "text"
      },
      "source": [
        "# Data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UnOhpJ-PR6HN",
        "colab_type": "text"
      },
      "source": [
        "## initial training dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-HWyGjobR4P_",
        "colab_type": "code",
        "outputId": "637786e7-2fd3-4087-c3ee-178d67493935",
        "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": 71
        }
      },
      "source": [
        "from google.colab import files\n",
        "\n",
        "uploaded = files.upload()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-8c5b0d87-fd08-4385-9155-97b28e74716f\" name=\"files[]\" multiple disabled />\n",
              "     <output id=\"result-8c5b0d87-fd08-4385-9155-97b28e74716f\">\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> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving photomask_trainingdata.zip to photomask_trainingdata.zip\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kjdIMZ9eSNXH",
        "colab_type": "code",
        "outputId": "0139372c-77a7-4e6f-f094-7131361afbf1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "!ls -la"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "total 3192\n",
            "drwxr-xr-x 1 root root    4096 Jun 28 12:17 .\n",
            "drwxr-xr-x 1 root root    4096 Jun 28 12:06 ..\n",
            "drwxr-xr-x 1 root root    4096 Jun 18 16:14 .config\n",
            "-rw-r--r-- 1 root root 3249054 Jun 28 12:18 photomask_trainingdata.zip\n",
            "drwxr-xr-x 1 root root    4096 Jun 18 16:14 sample_data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aqDN2K_7Tc-x",
        "colab_type": "code",
        "outputId": "fcce5ce2-4ff1-4325-c2da-96052295ed64",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "!unzip photomask_trainingdata.zip -d dataset"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  photomask_trainingdata.zip\n",
            "   creating: dataset/photomask_trainingdata/\n",
            "   creating: dataset/photomask_trainingdata/COMPLETE FEATURE/\n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/026_2.tif  \n",
            "   creating: dataset/__MACOSX/\n",
            "   creating: dataset/__MACOSX/photomask_trainingdata/\n",
            "   creating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/\n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._026_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/002_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._002_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/020_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._020_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/004_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._004_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/019_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._019_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/024_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._024_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/002_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._002_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/026_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._026_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/026_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._026_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/024_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._024_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/019_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._019_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/006_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._006_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/022_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._022_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/019_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._019_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/002_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._002_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/024_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._024_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/026_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._026_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/004_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._004_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/020_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._020_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/022_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._022_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/004_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._004_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/002_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._002_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/022_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._022_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/020_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._020_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/006_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._006_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/006_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._006_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/004_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._004_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/022_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._022_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/019_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._019_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/024_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._024_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/007_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._007_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/018_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._018_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/025_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._025_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/027_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._027_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/001_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._001_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/001_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._001_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/025_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._025_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/023_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._023_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/018_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._018_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/021_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._021_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/001_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._001_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/027_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._027_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/027_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._027_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/025_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._025_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/018_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._018_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/021_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._021_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/007_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._007_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/023_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._023_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/001_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._001_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/025_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._025_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/016.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._016.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/005_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._005_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/023_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._023_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/018_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._018_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/007_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._007_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/007_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._007_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/021_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._021_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/027_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._027_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/015.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._015.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/023_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._023_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/005_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._005_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/021_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._021_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/029_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._029_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/029_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._029_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/029_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._029_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/029_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._029_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/028_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._028_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/008_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._008_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/017_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._017_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/028_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._028_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/017_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._017_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/028_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._028_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/028_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._028_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/017_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._017_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/008_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._008_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/COMPLETE FEATURE/017_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/COMPLETE FEATURE/._017_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/._COMPLETE FEATURE  \n",
            "   creating: dataset/photomask_trainingdata/BORDERTOUCHER/\n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/006_1.tif  \n",
            "   creating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/\n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._006_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/020_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._020_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/011.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._011.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/010.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._010.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/012.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._012.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/005_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._005_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/005_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._005_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/030_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._030_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/030_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._030_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/030_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._030_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/009.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._009.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/030_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._030_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/031_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._031_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/008_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._008_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/033_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._033_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/008_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._008_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/031_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._031_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/033_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._033_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/031_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._031_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/033_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._033_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/033_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._033_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/BORDERTOUCHER/031_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/BORDERTOUCHER/._031_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/._BORDERTOUCHER  \n",
            "   creating: dataset/photomask_trainingdata/NO FEATURE/\n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/003_1.tif  \n",
            "   creating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/\n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._003_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/003_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._003_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/003_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._003_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/013.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._013.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/003_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._003_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/032_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._032_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/014_4.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._014_4.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/032_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._032_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/014_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._014_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/032_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._032_3.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/014_2.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._014_2.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/032_1.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._032_1.tif  \n",
            "  inflating: dataset/photomask_trainingdata/NO FEATURE/014_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/NO FEATURE/._014_3.tif  \n",
            "  inflating: dataset/__MACOSX/photomask_trainingdata/._NO FEATURE  \n",
            "  inflating: dataset/__MACOSX/._photomask_trainingdata  \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YyONXtKgTlJI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "rm -r dataset/__MACOSX/"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vg_2f1mcSpJA",
        "colab_type": "text"
      },
      "source": [
        "## TF Dataset based on the files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xb_gzYb2T38g",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "dataset_path = 'dataset/photomask_trainingdata/'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HtoOoFjETtDw",
        "colab_type": "code",
        "outputId": "c403c998-7106-4173-bf59-d01d0478c9bb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "!ls -la {dataset_path}"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "total 20\n",
            "drwxrwxrwx 5 root root 4096 Feb  5 17:17  .\n",
            "drwxr-xr-x 3 root root 4096 Jun 28 12:18  ..\n",
            "drwxrwxrwx 2 root root 4096 Feb  5 17:04  BORDERTOUCHER\n",
            "drwxrwxrwx 2 root root 4096 Feb  5 17:05 'COMPLETE FEATURE'\n",
            "drwxrwxrwx 2 root root 4096 Feb  5 17:04 'NO FEATURE'\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HxeXXIvySsLQ",
        "colab_type": "code",
        "outputId": "5d173b0a-bf7c-4f3f-9e6e-7af280ad52f2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "tf.version.VERSION"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'2.0.0-beta1'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ESlfgTKbVoqx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "work_image_size = [28, 28]\n",
        "\n",
        "batch_size = 16"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nxYSelu-WGqG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from os import listdir, path\n",
        "\n",
        "class_names = listdir(dataset_path)\n",
        "# print(class_names)\n",
        "\n",
        "class_folders = [(idx, path.join(dataset_path, class_name)) for idx, class_name in enumerate(class_names)]\n",
        "# print(class_folders)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rBRpk3phUyBm",
        "colab_type": "code",
        "outputId": "5728a295-d4af-4edf-a26e-68443cd36639",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "filenames = []\n",
        "class_ids = []\n",
        "for class_id, folder in class_folders:\n",
        "  files = listdir(folder)\n",
        "  \n",
        "  filenames.extend([path.join(folder, filename) for filename in listdir(folder)])\n",
        "  class_ids.extend([class_id for filename in listdir(folder)])\n",
        "\n",
        "\n",
        "print(len(filenames))\n",
        "assert len(class_ids) == len(class_ids)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "111\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wH2NMVREDUA7",
        "colab_type": "text"
      },
      "source": [
        "Unfortunately `tf.py_function` can only return a tuple. With generic `map` we can work with `dict` interface"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xnWV25GPuMGE",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from PIL import Image\n",
        "import numpy as np\n",
        "\n",
        "def parse_tiff_py(filename, class_id):\n",
        "    print(filename.numpy())\n",
        "    \n",
        "    im = Image.open(filename.numpy())\n",
        "    im = im.resize(work_image_size)\n",
        "    \n",
        "    return np.array(im), class_id"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hZ1NDRvMDseJ",
        "colab_type": "text"
      },
      "source": [
        "Now we just need to add our python function to `dataset` pipeline. We can do with follow `lambda` trick"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LOhA4Ym8U2Hz",
        "colab_type": "code",
        "outputId": "ec62371c-f5dc-40fa-d9d8-c76f83ca162c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "dataset = tf.data.Dataset.from_tensor_slices((filenames, class_ids))\n",
        "dataset = dataset.shuffle(len(filenames))\n",
        "\n",
        "dataset = dataset.map(lambda filename, label: tf.py_function(\n",
        "              parse_tiff_py, \n",
        "              (filename, label),  \n",
        "              [tf.uint8, label.dtype])\n",
        "          )\n",
        "\n",
        "dataset"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<MapDataset shapes: (<unknown>, <unknown>), types: (tf.uint8, tf.int32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rDG_Sg5eZNjM",
        "colab_type": "code",
        "outputId": "0f916179-c573-4465-8fa8-0d46c1c17061",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "for row in dataset.take(1):\n",
        "  print(row)\n",
        "  image = row[0]\n",
        "  label = row[1]\n",
        "  \n",
        "  print(image.shape)"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "b'dataset/photomask_trainingdata/COMPLETE FEATURE/024_2.tif'\n",
            "(<tf.Tensor: id=216, shape=(28, 28), dtype=uint8, numpy=\n",
            "array([[ 33,  18,  75,  51,  39,  53,  96,  94,  70,  86,  59,  27,  95,\n",
            "         41,  48,  71,  96,  80,  71,  60,  71,  61,  75,  61,  38,  42,\n",
            "         40,  14],\n",
            "       [ 79,  30,  58,  36,  40,  35,  39,  40,  82,  39,  72,  24,  45,\n",
            "         52,  48,  93,  22,  77,  48,  29,  97,  43,  81,  42,  66,  37,\n",
            "         65,  81],\n",
            "       [ 61,  44,  62,  61,  50,  65,  86,  41,  16, 103, 101,  73,  78,\n",
            "         55,  39,  32,  70,  69,  46,  50,  19,  36,  53,  39,  65,  23,\n",
            "         41,  67],\n",
            "       [ 41,  65,  52,  18,  64,  38,  70,  46,  26,  59,  50,  48,  98,\n",
            "         60,  66,  84, 114,  44,  84,  73,  59,  63,  31,  23,  44,  30,\n",
            "         50,  48],\n",
            "       [ 35,  16,  40,  90,  25,  34,  32,  77,  46,  49, 124,  99,  72,\n",
            "         52,  87,  13,  83,  43,  41,  70,  67,  70,  83,  64,  97,  35,\n",
            "         49,  41],\n",
            "       [ 82,  66,  44,  30,  38,  49,  40,  57,  45,  24,  55,  37,  72,\n",
            "         61,  59,  64,  53,  39,  75,  21,  43,  47,  30,  43,  27,  29,\n",
            "         61,  46],\n",
            "       [ 40,  47,  96,  21,  67,  20,  40,  81,  40,  59,  27,  29,  68,\n",
            "         54,  48,  67,  48,  48,  58,  60, 109,  86,  27,  59,  44,  39,\n",
            "         35,  53],\n",
            "       [ 38,  51,  44,  29,  50,  67,  52,  76,   9,  69,  73,  37,  71,\n",
            "         77,  40,  55,  45, 102,  96,  52,  59,  33,  13,  76,  36,  34,\n",
            "         88,  48],\n",
            "       [ 51,  55,  29,  63,  55,  74,  91,  63,  85,  66,  31,  69,  80,\n",
            "         54,  79,  70,  76,  47,  82,  66,   6,  62, 119,  70,  74,  34,\n",
            "         59,  40],\n",
            "       [ 96,  66,  71,  44,  22,  52,  12,  77,  83,  22,  83,  28,  28,\n",
            "         80,  88,  80,  84,  38,  54,  75,  38,  53,  90,  55,  70,  30,\n",
            "         49,  48],\n",
            "       [ 81,  83,  57,  67,  52,  46,  48,  74,  77,  70,  34,  95,  87,\n",
            "        112,  99, 120,  24,  65,  66,  54, 104,  78,  83, 101,  57,  45,\n",
            "         72,  92],\n",
            "       [ 46,  51,  34,  30,  83,  67,  75,  47,  42,  94, 111, 161, 176,\n",
            "        161, 169,  75, 121,  44,  53,  25,  61,  83,  80,  69,  78,  76,\n",
            "         66,  89],\n",
            "       [ 51,  50,  66,  74,  56,  71,  85,  17,  42,  62,  87, 117, 183,\n",
            "        199, 113, 113,  69,  84,  47,  51,  32,  32,  26,  39,  63,  77,\n",
            "         35,  70],\n",
            "       [ 61,  99,  82,  89,  19,  75,  95,  64,   9,  30,  84,  75, 140,\n",
            "        151, 173,  85, 134,  51,  81,  20,  56,  85,  69,  58,  70,  93,\n",
            "         48,  29],\n",
            "       [ 43,  35,  26,  67,  56,  34,  49,  32,  62, 110, 146, 181, 136,\n",
            "        138, 161, 171, 101,  34,  65,  54,  39,  57,  95, 108,  17, 104,\n",
            "         63,  75],\n",
            "       [ 32,  84,  45, 111,  53, 107,  51,  45,  41,  66, 111, 178,  73,\n",
            "        104, 119, 126,   8,  50,  43,  24,  53,  60,  79,  49,  45,  46,\n",
            "         37,  56],\n",
            "       [ 61,  50,  41,  69,  72,  59,  30,  47,  15,  64,  88, 109,  60,\n",
            "         71,  65,  17,  37,  33,  12,  26,  34,  64,  42,  61,  32,  41,\n",
            "         31,  55],\n",
            "       [ 54,  50,  86,  71,  79,  76,  24,  35,  26,  33,  85,  30,  24,\n",
            "         47,  53,  59,   0,   2,  66,  51,  42,  67,  62,  44,  41,  78,\n",
            "         50,  72],\n",
            "       [139,  56,  52, 116,  43,  82,  86,  44,  33,  43,  43,  54,  61,\n",
            "         22,  67,  41,  44,  42,  37,  38,  56,   0,  60,  88,  46,  25,\n",
            "         85,  49],\n",
            "       [ 67,  29,  35,  51,  69,  17,  50,  54,  80,  38,  40,  86,  42,\n",
            "         31,  51,   2,  63,  53,  73,  32,  51,  35,  39,  80, 121,  63,\n",
            "         39,  70],\n",
            "       [100,  30,  19,  49,  60,  55,  51,  81,  76,  46,  68,  42,  58,\n",
            "         36,  32,  30, 102,  82,   7,  48,  37,  14,  58,  20,  74,  45,\n",
            "         12,  76],\n",
            "       [114,  71,  57,  65,  65,  61,  47,  20,  41,  92,  27,  60,   8,\n",
            "         88,  42,  37,  72,  10,  79,  34,  45,  31,  54,  47,  27,  94,\n",
            "         48,  65],\n",
            "       [ 79,  43,  42,  32,  74,  91,  23,  64,  66,  84,  98,  87,  72,\n",
            "         50, 106,  15,  52,  83,  73,  26,  33,  87,  51,  22,  14,  94,\n",
            "        105,  76],\n",
            "       [101,  48,  40,  82,  42,  88,  67,  43, 106,  69,  58,  73,  65,\n",
            "         45,  57,  53,  42, 115,  94,  37,  73,  60,  83,  47,  50,  63,\n",
            "         82,  64],\n",
            "       [ 70,  32,  51,  83,  40,  43,  71,  87,  27,  42,  28,  57, 118,\n",
            "         84,  65,  79,  31,  83,  18,   7,  89,  88,  79,  63,  11,  55,\n",
            "         55,  98],\n",
            "       [  9,  51,  84,  74,  84,  31, 101,  63,  49,  75,  80,  54,  43,\n",
            "         21,  51,  12,  53,  37,  40,  95,  64,  78,  65,  40, 105,  30,\n",
            "         89,  43],\n",
            "       [ 52, 100,  14,  58,  19,  68,  52,  95,  59,  93,  54,  63,  52,\n",
            "         56,  61,  32,  76,  85,  54,  66,  92,  81,  44,  33,  68,  58,\n",
            "        103,  83],\n",
            "       [ 48,  43, 100,  49, 110,  18,  70,  35, 137,  59,  52,  90,  44,\n",
            "        124,  61,  57,  98,  57,  75,  60,  82,  49,  37,  57,  41,  43,\n",
            "         57,  79]], dtype=uint8)>, <tf.Tensor: id=217, shape=(), dtype=int32, numpy=2>)\n",
            "(28, 28)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zl0cR3ZWuoGy",
        "colab_type": "code",
        "outputId": "671e649a-3d85-4a7b-a6fd-7c311d38627f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        }
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "print(image.shape)\n",
        "plt.imshow(image, cmap=plt.get_cmap(\"gray\"))\n",
        "\n",
        "print(f\"Label - {class_names[label]}\")"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(28, 28)\n",
            "Label - BORDERTOUCHER\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhcgUmV+ITJH5hcgUmV+ITPk/CIGSRWiJHGQAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YnXafs9lEjlJ",
        "colab_type": "text"
      },
      "source": [
        "# TemsorFlow Addons"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PXffDK7_Eo5C",
        "colab_type": "code",
        "outputId": "43682968-ef36-4b92-ca88-28c0e1b253d7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 752
        }
      },
      "source": [
        "!pip install tensorflow-io-2.0-preview"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting tensorflow-io-2.0-preview\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/2b/f5/e89140c8a1f50c075e3d3458a54846d915164ba5b6dda7c8b58b34ca456b/tensorflow_io_2.0_preview-0.7.0.dev1107-cp36-cp36m-manylinux1_x86_64.whl (54.4MB)\n",
            "\u001b[K     |████████████████████████████████| 54.4MB 1.9MB/s \n",
            "\u001b[?25hCollecting tensorflow==2.0.0b1 (from tensorflow-io-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/29/6c/2c9a5c4d095c63c2fb37d20def0e4f92685f7aee9243d6aae25862694fd1/tensorflow-2.0.0b1-cp36-cp36m-manylinux1_x86_64.whl (87.9MB)\n",
            "\u001b[K     |████████████████████████████████| 87.9MB 30.0MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy<2.0,>=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.16.4)\n",
            "Requirement already satisfied: google-pasta>=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.1.7)\n",
            "Requirement already satisfied: wrapt>=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.11.1)\n",
            "Requirement already satisfied: grpcio>=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.15.0)\n",
            "Requirement already satisfied: astor>=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.8.0)\n",
            "Requirement already satisfied: keras-applications>=1.0.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.0.8)\n",
            "Collecting tb-nightly<1.14.0a20190604,>=1.14.0a20190603 (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/a4/96/571b875cd81dda9d5dfa1422a4f9d749e67c0a8d4f4f0b33a4e5f5f35e27/tb_nightly-1.14.0a20190603-py3-none-any.whl (3.1MB)\n",
            "\u001b[K     |████████████████████████████████| 3.1MB 30.5MB/s \n",
            "\u001b[?25hRequirement already satisfied: keras-preprocessing>=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.1.0)\n",
            "Requirement already satisfied: gast>=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.2.2)\n",
            "Requirement already satisfied: protobuf>=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (3.7.1)\n",
            "Collecting tf-estimator-nightly<1.14.0.dev2019060502,>=1.14.0.dev2019060501 (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/32/dd/99c47dd007dcf10d63fd895611b063732646f23059c618a373e85019eb0e/tf_estimator_nightly-1.14.0.dev2019060501-py2.py3-none-any.whl (496kB)\n",
            "\u001b[K     |████████████████████████████████| 501kB 42.0MB/s \n",
            "\u001b[?25hRequirement already satisfied: termcolor>=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.1.0)\n",
            "Requirement already satisfied: six>=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (1.12.0)\n",
            "Requirement already satisfied: wheel>=0.26 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.33.4)\n",
            "Requirement already satisfied: absl-py>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.7.1)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications>=1.0.6->tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (2.8.0)\n",
            "Requirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tb-nightly<1.14.0a20190604,>=1.14.0a20190603->tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (3.1.1)\n",
            "Requirement already satisfied: setuptools>=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tb-nightly<1.14.0a20190604,>=1.14.0a20190603->tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (41.0.1)\n",
            "Requirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tb-nightly<1.14.0a20190604,>=1.14.0a20190603->tensorflow==2.0.0b1->tensorflow-io-2.0-preview) (0.15.4)\n",
            "Installing collected packages: tb-nightly, tf-estimator-nightly, tensorflow, tensorflow-io-2.0-preview\n",
            "  Found existing installation: tb-nightly 1.14.0a20190614\n",
            "    Uninstalling tb-nightly-1.14.0a20190614:\n",
            "      Successfully uninstalled tb-nightly-1.14.0a20190614\n",
            "  Found existing installation: tensorflow 1.14.0rc1\n",
            "    Uninstalling tensorflow-1.14.0rc1:\n",
            "      Successfully uninstalled tensorflow-1.14.0rc1\n",
            "Successfully installed tb-nightly-1.14.0a20190603 tensorflow-2.0.0b1 tensorflow-io-2.0-preview-0.7.0.dev1107 tf-estimator-nightly-1.14.0.dev2019060501\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.colab-display-data+json": {
              "pip_warning": {
                "packages": [
                  "tensorboard",
                  "tensorflow",
                  "tensorflow_estimator"
                ]
              }
            }
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LrCRUvMtFeFm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow_io.image import TIFFDataset"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TtiH6wW0GAxY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "dataset = TIFFDataset(filenames)\n",
        "dataset = dataset.shuffle(len(filenames))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hldWEvD6CBcH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "8eff4c40-66b3-4272-f03e-3189fc379468"
      },
      "source": [
        "for image in dataset.take(4):\n",
        "  print(image.shape)"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(131, 131, 4)\n",
            "(194, 194, 4)\n",
            "(123, 123, 4)\n",
            "(86, 86, 4)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dz0Od3TPTUOw",
        "colab_type": "code",
        "outputId": "ab7a2130-9a9f-4414-9f37-7a7db6410a59",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        }
      },
      "source": [
        "print(image.shape)\n",
        "plt.imshow(image.numpy(), cmap=plt.get_cmap(\"gray\"))"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(86, 86, 4)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f90b76aae80>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 69
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n1n74zZpNQ2s//GbNpqG1H36zZtPQ/j+yyTbHdEf7/QAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QLiio1VcJbX0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}