N3_GAN_CelebA_final.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "N3_GAN_CelebA_final.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "H7dzjjdi826X",
        "EffqIllQ826X",
        "162Xn5Ut826j",
        "AnPwn7lC826t",
        "vHaosi-7ebvp",
        "XDrmPOFuhp0t",
        "CceRl5Fp827G",
        "4HamjGt0827M"
      ]
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "pzIk_n8ZKaL_"
      },
      "source": [
        "# 1 ) Data Preprocessing and Exploring\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "2Bkr-BRO96D_"
      },
      "source": [
        "### Data Download"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "dbxG-T0PSWn7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 548
        },
        "outputId": "f34cca29-36be-44ce-9407-245a9cc9d2a7"
      },
      "source": [
        "# Download the data\n",
        "from data_loading import folder_download\n",
        "!wget -qnc https://github.com/segnolin/google-drive-folder-downloader/raw/master/download.py\n",
        "# Download the data\n",
        "folder_id = \"0B7EVK8r0v71pWEZsZE9oNnFzTm8\"\n",
        "folder_download(folder_id=folder_id)\n",
        "\n",
        "% cd /content/CelebA/Img/\n",
        "!unzip img_align_celeba.zip"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py\u001b[0m in \u001b[0;36m_input_request\u001b[0;34m(self, prompt, ident, parent, password)\u001b[0m\n\u001b[1;32m    728\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 729\u001b[0;31m                 \u001b[0mident\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreply\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrecv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstdin_socket\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    730\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/jupyter_client/session.py\u001b[0m in \u001b[0;36mrecv\u001b[0;34m(self, socket, mode, content, copy)\u001b[0m\n\u001b[1;32m    802\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 803\u001b[0;31m             \u001b[0mmsg_list\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msocket\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrecv_multipart\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmode\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    804\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mzmq\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mZMQError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/zmq/sugar/socket.py\u001b[0m in \u001b[0;36mrecv_multipart\u001b[0;34m(self, flags, copy, track)\u001b[0m\n\u001b[1;32m    474\u001b[0m         \"\"\"\n\u001b[0;32m--> 475\u001b[0;31m         \u001b[0mparts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrecv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mflags\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrack\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtrack\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    476\u001b[0m         \u001b[0;31m# have first part already, only loop while more to receive\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32mzmq/backend/cython/socket.pyx\u001b[0m in \u001b[0;36mzmq.backend.cython.socket.Socket.recv\u001b[0;34m()\u001b[0m\n",
            "\u001b[0;32mzmq/backend/cython/socket.pyx\u001b[0m in \u001b[0;36mzmq.backend.cython.socket.Socket.recv\u001b[0;34m()\u001b[0m\n",
            "\u001b[0;32mzmq/backend/cython/socket.pyx\u001b[0m in \u001b[0;36mzmq.backend.cython.socket._recv_copy\u001b[0;34m()\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/zmq/backend/cython/checkrc.pxd\u001b[0m in \u001b[0;36mzmq.backend.cython.checkrc._check_rc\u001b[0;34m()\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: ",
            "\nDuring handling of the above exception, another exception occurred:\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-4-845c2663e501>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m# Download the data\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mfolder_id\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"0B7EVK8r0v71pWEZsZE9oNnFzTm8\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 6\u001b[0;31m \u001b[0mfolder_download\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfolder_id\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfolder_id\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmagic\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'cd /content/CelebA/Img/'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/content/data_loading.py\u001b[0m in \u001b[0;36mfolder_download\u001b[0;34m(folder_id)\u001b[0m\n\u001b[1;32m     19\u001b[0m   \u001b[0;31m# authenticate\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     20\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 21\u001b[0;31m   \u001b[0mauth\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mauthenticate_user\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     22\u001b[0m   \u001b[0;31m# get folder_name\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     23\u001b[0m   \u001b[0mservice\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mbuild\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'drive'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'v3'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/google/colab/auth.py\u001b[0m in \u001b[0;36mauthenticate_user\u001b[0;34m(clear_output)\u001b[0m\n\u001b[1;32m    145\u001b[0m     \u001b[0mcontext_manager\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_output\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtemporary\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mclear_output\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0m_noop\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    146\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mcontext_manager\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 147\u001b[0;31m       \u001b[0m_gcloud_login\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    148\u001b[0m     \u001b[0m_install_adc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    149\u001b[0m     \u001b[0mcolab_tpu_addr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_os\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0menviron\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'COLAB_TPU_ADDR'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m''\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/google/colab/auth.py\u001b[0m in \u001b[0;36m_gcloud_login\u001b[0;34m()\u001b[0m\n\u001b[1;32m     89\u001b[0m     \u001b[0;31m# https://github.com/jupyter/notebook/issues/3159\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     90\u001b[0m     \u001b[0mprompt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mprompt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrstrip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 91\u001b[0;31m     \u001b[0mcode\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_getpass\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgetpass\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mprompt\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m' '\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     92\u001b[0m     \u001b[0mgcloud_process\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcommunicate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcode\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstrip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     93\u001b[0m   \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py\u001b[0m in \u001b[0;36mgetpass\u001b[0;34m(self, prompt, stream)\u001b[0m\n\u001b[1;32m    685\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_parent_ident\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    686\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_parent_header\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 687\u001b[0;31m             \u001b[0mpassword\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    688\u001b[0m         )\n\u001b[1;32m    689\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py\u001b[0m in \u001b[0;36m_input_request\u001b[0;34m(self, prompt, ident, parent, password)\u001b[0m\n\u001b[1;32m    732\u001b[0m             \u001b[0;32mexcept\u001b[0m \u001b[0mKeyboardInterrupt\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    733\u001b[0m                 \u001b[0;31m# re-raise KeyboardInterrupt, to truncate traceback\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 734\u001b[0;31m                 \u001b[0;32mraise\u001b[0m \u001b[0mKeyboardInterrupt\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    735\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    736\u001b[0m                 \u001b[0;32mbreak\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "wWQCigycaLBQ"
      },
      "source": [
        "## Imports"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "SuwogZZMImXC",
        "colab": {}
      },
      "source": [
        "import argparse\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import colorsys\n",
        "import os\n",
        "import data_loading\n",
        "from data_loading import Data_Generation, Load_Data\n",
        "import model\n",
        "import train\n",
        "\n",
        "parser = argparse.ArgumentParser()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "sayXv8Mr-N6Q",
        "colab": {}
      },
      "source": [
        "parser = argparse.ArgumentParser()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "uWlGyU_-x_OB",
        "colab": {}
      },
      "source": [
        "# create directory to be used in the codes.\n",
        "\n",
        "!mkdir /content/CelebA/checkpoints\n",
        "!mkdir /content/CelebA/categorical\n",
        "!mkdir /content/CelebA/training_log\n",
        "!mkdir /content/CelebA/images\n",
        "\n",
        "!mkdir /content/CelebA/images/full\n",
        "!mkdir /content/CelebA/images/classes\n",
        "!mkdir /content/CelebA/images/disrc\n",
        "!mkdir /content/CelebA/images/random"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "OkDJhH1V2TZ6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "ba189b69-846c-4c09-b9d2-fb7ec9e101f6"
      },
      "source": [
        "# Move the list_attr_celeba_2.csv to Anno\n",
        "import shutil\n",
        "shutil.move(\"/content/list_attr_celeba_2.csv\", \"/content/CelebA/Anno/\") \n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'/content/CelebA/Anno/list_attr_celeba_2.csv'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "P1AJGQ2tddw8",
        "colab": {}
      },
      "source": [
        "parser.add_argument('--CSVPATH',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/Anno/list_attr_celeba_2.csv\",\n",
        "                    help = \"path to file of list attribute\")\n",
        "\n",
        "parser.add_argument('--CHECKPOINTS',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/checkpoints\",\n",
        "                    help = \"path to save model checkpoints\")\n",
        "\n",
        "parser.add_argument('--NUMPY_FILES',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/categorical\",\n",
        "                    help = \"path to numpy saved images and labels as .npy files in the range [-1;1]\")\n",
        "\n",
        "parser.add_argument('--LOG',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/training_log\",\n",
        "                    help = \"path to training logs\") \n",
        "\n",
        "parser.add_argument('--IMAGES',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images\",\n",
        "                    help = \"directory of all generated images during training/testing.\") \n",
        "\n",
        "\n",
        "parser.add_argument('--FULL_SAVE',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images/full\"\n",
        "                   ) \n",
        "\n",
        "parser.add_argument('--CLASSES_SAVE',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images/classes\"\n",
        "                   ) \n",
        "\n",
        "parser.add_argument('--DISCR_SAVE',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images/disrc\"\n",
        "                    ) \n",
        "\n",
        "parser.add_argument('--RANDOM_SAVE',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/Images/random\"\n",
        "                    ) \n",
        "\n",
        "parser.add_argument('--ATTRIBUTE_HISTOGRAM',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images/attributes_histo.png\",\n",
        "                    help = \"path to save attribute histrogram image\") \n",
        "\n",
        "parser.add_argument('--ATTRIBUTE_DISTRIBUTION',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/images/attributes_distr.png\",\n",
        "                    help = \"path to save attribute distribution image\") \n",
        "\n",
        "parser.add_argument('--DATA_PATH',\n",
        "                    type=str,\n",
        "                    default = \"/content/CelebA/Img/img_align_celeba\",\n",
        "                    help = \"path to the actual images from the celeba dataset\")\n",
        "\n",
        "parser.add_argument('--IMAGE_SIZE',\n",
        "                    type=int,\n",
        "                    default = 128,\n",
        "                    help = \"size of the image\") \n",
        "\n",
        "parser.add_argument('--NUMBER_OF_CLASSES',\n",
        "                    type=int,\n",
        "                    default = 3, \n",
        "                    help = \"total number of classes\")\n",
        "\n",
        "parser.add_argument('--BATCH_SIZE',\n",
        "                    type=int,\n",
        "                    default = 128, \n",
        "                    help = \"batch size\") \n",
        "args, _ = parser.parse_known_args()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "leYhA-xT1ksL",
        "colab": {}
      },
      "source": [
        "# read the data attribute file\n",
        "df = pd.read_csv(args.CSVPATH, delimiter=\";\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "H7dzjjdi826X"
      },
      "source": [
        "## Exploration"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "EffqIllQ826X"
      },
      "source": [
        "### How many attributes / picture on average"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "eLlhnGV9826X",
        "colab": {}
      },
      "source": [
        "no_fn = df.drop(\"filename\", axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "gB6OdH2M826Z",
        "colab": {}
      },
      "source": [
        "avg = no_fn.add(1,axis=0).sum(axis=1).apply(lambda x: x//2).aggregate(\"average\")\n",
        "std = no_fn.add(1,axis=0).sum(axis=1).apply(lambda x: x//2).aggregate(np.std)\n",
        "df_min = no_fn.add(1,axis=0).sum(axis=1).apply(lambda x: x//2).aggregate(np.min)\n",
        "df_max = no_fn.add(1,axis=0).sum(axis=1).apply(lambda x: x//2).aggregate(np.max)\n",
        "counts_series = no_fn.add(1,axis=0).sum(axis=1).apply(lambda x: x//2)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "K7l3D0e1826c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        },
        "outputId": "cf2dc2b7-0464-4963-b2f3-41868750f3b2"
      },
      "source": [
        "# Get the Indices of the highest and lowest amount of attributes per person\n",
        "low_index = counts_series[counts_series==1].index[0]\n",
        "high_index = counts_series[counts_series==20].index[0]\n",
        "\n",
        "# Get the corresponding image-filenames\n",
        "print(f\"Only one attribute: \\t{df.iloc[low_index].filename}\")\n",
        "print(f\"Twenty attribute: \\t{df.iloc[high_index].filename}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Only one attribute: \t003259.jpg\n",
            "Twenty attribute: \t014308.jpg\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "7N8KbdYx826f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 84
        },
        "outputId": "3e0a3bcb-aa41-4b01-d3a1-7b815d0bc747"
      },
      "source": [
        "print(f\"Average of Attributes/Person: \\t{np.round(avg, decimals = 2)}\")\n",
        "print(f\"Std of Attributes/Person: \\t{np.round(std, decimals = 2)}\")\n",
        "print(f\"Lowest amount of attributes: \\t{np.round(df_min, decimals = 2)}\")\n",
        "print(f\"Highest amount of attributes: \\t{np.round(df_max, decimals = 2)}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average of Attributes/Person: \t9.03\n",
            "Std of Attributes/Person: \t2.94\n",
            "Lowest amount of attributes: \t1\n",
            "Highest amount of attributes: \t20\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "YxF999AE-Lyu",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 382
        },
        "outputId": "774eb835-9b2c-4a1a-91c7-39cd547d4d47"
      },
      "source": [
        "fig = plt.figure(figsize=(7,4),dpi=100)\n",
        "plt.rc('axes', axisbelow=True)\n",
        "plt.hist(counts_series, bins=20, color = \"#900d0d\", edgecolor= \"black\", range=(0,20))\n",
        "plt.grid()\n",
        "plt.xlabel(\"Amount of Positive Attributes per Person\")\n",
        "plt.ylabel(\"Occurrences\")\n",
        "plt.savefig(args.ATTRIBUTE_HISTOGRAM, dpi=300)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 700x400 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "162Xn5Ut826j"
      },
      "source": [
        "### Distribution of Attributes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "uzFdCz9j826m",
        "colab": {}
      },
      "source": [
        "no_fn = df.drop(\"filename\", axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5bmo-r7N826n",
        "colab": {}
      },
      "source": [
        "totals = no_fn.add(1,axis=0).sum(axis=0)//2"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "9HOJfJgq826p",
        "colab": {}
      },
      "source": [
        "value = list(np.linspace(0.3,0.8, num=40))\n",
        "N = 40\n",
        "HSV_tuples = [(0.01, 0.9, value[x]) for x in range(N)]\n",
        "RGB_tuples = map(lambda x: colorsys.hsv_to_rgb(*x), HSV_tuples)\n",
        "list_rgb = list(RGB_tuples)[::-1]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "zh-_BZKJ826r",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 407
        },
        "outputId": "5bd7e5e7-9b76-4c62-ec33-a1c88468f33b"
      },
      "source": [
        "fig = plt.figure(figsize=(9,4),dpi=100)\n",
        "\n",
        "plt.rc('axes', axisbelow=True)\n",
        "plt.grid()\n",
        "barlist = plt.bar(totals.sort_values(ascending=False).index, totals.sort_values(ascending=False), align=\"center\", edgecolor=\"black\")\n",
        "plt.xticks(rotation='vertical')\n",
        "for i in range(len(barlist)):\n",
        "    barlist[i].set_color(list_rgb[i])\n",
        "    barlist[i].set_edgecolor(\"black\")\n",
        "plt.ylabel(\"Total Occurrences\")\n",
        "plt.tight_layout()\n",
        "plt.savefig(args.ATTRIBUTE_DISTRIBUTION, dpi=300)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 900x400 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "AnPwn7lC826t"
      },
      "source": [
        "## Create numpy files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "-nmxGUwVDcYQ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 168
        },
        "outputId": "8377d0cb-fe05-415a-c86d-f7a0d577b3f4"
      },
      "source": [
        "Data_Generation(args, df)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Attribute Eyeglasses found 12000 examples.\n",
            "Looking for file: /content/CelebA/categorical/training_data_128_128_Eyeglasses_12000.npy\n",
            "Loading previous training pickle...\n",
            "Attribute Rosy_Cheeks found 12000 examples.\n",
            "Looking for file: /content/CelebA/categorical/training_data_128_128_Rosy_Cheeks_12000.npy\n",
            "Loading previous training pickle...\n",
            "Attribute Goatee found 12000 examples.\n",
            "Looking for file: /content/CelebA/categorical/training_data_128_128_Goatee_12000.npy\n",
            "Loading previous training pickle...\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "vHaosi-7ebvp"
      },
      "source": [
        "# 2) Conditional DCGAN"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "XDrmPOFuhp0t"
      },
      "source": [
        "## Loading the Data - conditional GAN\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "ridcPnM_IUJI",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e2af4df9-44d1-4b91-abdd-5c73c84e53e3"
      },
      "source": [
        "X.shape, y.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((36000, 128, 128, 3), (36000,))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "VSq94CsJH3xz",
        "colab": {}
      },
      "source": [
        "X, y = Load_Data(args)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "hek2y1ZthsE-",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "ABMOYw_bhxYN"
      },
      "source": [
        "## Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "eZTTtKRunqpL",
        "colab": {}
      },
      "source": [
        "# initialize models\n",
        "# size of the latent space\n",
        "latent_dim = 100\n",
        "# create the discriminator\n",
        "d_model = model.define_cond_discriminator(args)\n",
        "# create the generator\n",
        "g_model = model.define_cond_generator(100, args)\n",
        "# create the gan\n",
        "gan_model = model.define_cond_gan(g_model, d_model)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "uMRGdu54ORLa",
        "colab": {}
      },
      "source": [
        "train.Training(g_model, d_model, gan_model, X, y, latent_dim, args, n_epochs=10000, prefix=\"final_\", epoch_start=0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "E5DSjEK_827A",
        "colab": {}
      },
      "source": [
        "# ONLY here is the prefix not ending with _ due to the pattern matching process of finding the right file\n",
        "#n_epoch = load_latest_model(d_model, g_model, os.path.join(FACEDATADIR, \"checkpoints\"), prefix=\"full40_cDCGAN\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "qyWoG9hm827C",
        "colab": {}
      },
      "source": [
        "# If we already have some pretrained weights, load it\n",
        "import glob\n",
        "#epoch =  Epich number\n",
        "\n",
        "# check if the model already exists.\n",
        "# if it exist, that means we have already trained it before, \n",
        "# and will use these weights, it will start training from where we left off.\n",
        "PATH_MODEL = os.path.join(args.CHECKPOINTS, \"*.h5\")\n",
        "\n",
        "if glob.glob(PATH_MODEL):\n",
        "  d_model.load_weights(os.path.join(args.CHECKPOINTS,f\"final_cgan_discriminator_epoch_{epoch}.h5\"))\n",
        "  g_model.load_weights(os.path.join(args.CHECKPOINTS,f\"final_cgan_generator_epoch_{epoch}.h5\"))\n",
        "  epoch = epoch + 1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "r5pya_neo8b9",
        "scrolled": true,
        "colab": {}
      },
      "source": [
        "train(g_model, d_model, gan_model, X,y, latent_dim, n_epochs=10000, n_batch=128, prefix=\"final_\", n_classes=3, epoch_start=0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "CceRl5Fp827G"
      },
      "source": [
        "# 3) Visualize model performance"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "QsDnH1KA827G"
      },
      "source": [
        "## Loss"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "1hzQfcTS827G",
        "colab": {}
      },
      "source": [
        "training_log = pd.read_csv(os.path.join(LOG,\"final_training.csv\"), delimiter=\";\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "uIwEMFkl827I",
        "colab": {}
      },
      "source": [
        "training_log.head()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5pAO98Ik827K",
        "colab": {}
      },
      "source": [
        "fig = plt.figure(figsize=(15,7))\n",
        "plt.grid()\n",
        "plt.plot(training_log[\" d_loss_real\"], c=\"#900d0d\", label=\"D loss real\", linewidth=3, alpha=0.85)\n",
        "plt.plot(training_log[\" d_loss_fake\"], c=\"#cf1b1b\", label=\"D loss fake\", linewidth=3, alpha=0.85)\n",
        "plt.plot(training_log[\" g_loss\"], c=\"#206a5d\", label=\"G loss\", linewidth=3, alpha=0.85)\n",
        "plt.legend()\n",
        "plt.title(\"Losses over Epochs (Binary Cross Entropy)\")\n",
        "plt.xticks([0,10,20,30,40,50,60,70,80,90,95])\n",
        "plt.savefig(os.path.join(LOG,\"final_training_fig.png\"), dpi=300)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "4HamjGt0827M"
      },
      "source": [
        "## Save randomly generated pictures"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "KnMdjlle827M",
        "colab": {}
      },
      "source": [
        "[fake_pic, fake_label], _ = generate_fake_samples(g_model, 100, 300) # 300 instances"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "mVbUu9Rq827O",
        "colab": {}
      },
      "source": [
        "for i in range(300):\n",
        "    # rescale to 0-255, reshape and cast type \n",
        "    pic_data = (((fake_pic[i]+1)/2).reshape(128,128,3)*255).astype(np.uint8)\n",
        "    pic = Image.fromarray(pic_data)\n",
        "    pic.save(os.path.join(RANDOM_SAVE,f\"{CLASS_HEADER[fake_label[i]]}_{i}.png\"))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "S_wcRye-827Q"
      },
      "source": [
        "## [Explore latent space](https://machinelearningmastery.com/how-to-interpolate-and-perform-vector-arithmetic-with-faces-using-a-generative-adversarial-network/)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "QZVtNSUD827Q",
        "colab": {}
      },
      "source": [
        "# uniform interpolation between two points in latent space\n",
        "def interpolate_points(p1, p2, n_steps=10):\n",
        "    # interpolate ratios between the points\n",
        "    ratios = np.linspace(0, 1, num=n_steps)\n",
        "    # linear interpolate vectors\n",
        "    vectors = list()\n",
        "    for ratio in ratios:\n",
        "        v = (1.0 - ratio) * p1 + ratio * p2\n",
        "        vectors.append(v)\n",
        "    #print(vectors)\n",
        "    return np.stack(vectors)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5Zgoh_2_827S",
        "colab": {}
      },
      "source": [
        "# generate points in latent space\n",
        "n = 20\n",
        "pts = generate_latent_points(100, n)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "9ZAo8Yvp827U",
        "colab": {}
      },
      "source": [
        "interpolated = interpolate_points(pts[0][1], pts[0][1+1])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "BIJ94KWJ827X",
        "colab": {},
        "outputId": "481e8b36-274b-4de8-feb9-679db0c6572a"
      },
      "source": [
        "interpolated.shape, (np.ones((10,))*pts[1][i])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((10, 100), (10,))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "8l5D1HE7827Z",
        "colab": {}
      },
      "source": [
        "# uniform interpolation between two points in latent space\n",
        "def interpolate_points(p1, p2, n_steps=10):\n",
        "    # interpolate ratios between the points\n",
        "    ratios = np.linspace(0, 1, num=n_steps)\n",
        "    # linear interpolate vectors\n",
        "    vectors = list()\n",
        "    for ratio in ratios:\n",
        "        v = (1.0 - ratio) * p1 + ratio * p2\n",
        "        vectors.append(v)\n",
        "    #print(vectors)\n",
        "    return np.stack(vectors)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pfRXyB6ko7lO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def save_latent_interpolations(path, n):\n",
        "    for i in range(n):\n",
        "        # generate points in latent space\n",
        "        n = 6\n",
        "        pts = generate_latent_points(100, n)\n",
        "\n",
        "        # interpolate pairs\n",
        "        results = None\n",
        "        class_label = 0\n",
        "        for i in range(0, n, 2):\n",
        "            # interpolate points in latent space\n",
        "            interpolated = interpolate_points(pts[0][i], pts[0][i+1])\n",
        "            # generate images\n",
        "            X = g_model.predict([interpolated, np.ones((10,))*class_label])\n",
        "            class_label = class_label + 1\n",
        "            # scale from [-1,1] to [0,1]\n",
        "            X = (X + 1) / 2.0\n",
        "            if results is None:\n",
        "                results = X\n",
        "            else:\n",
        "                results = np.vstack((results, X))\n",
        "        # plot the result\n",
        "        #plot_generated(results, 10)\n",
        "\n",
        "        fig, rows = plt.subplots(nrows=3, ncols=10, figsize=(25,7))\n",
        "        i = 0\n",
        "        for row in rows:\n",
        "            for ax in row:\n",
        "                ax.imshow(results[i])\n",
        "                ax.axis(\"off\")\n",
        "                i=i+1\n",
        "\n",
        "        #plt.tight_layout(rect=[0, 0.03, 1, 0.95])\n",
        "        plt.tight_layout()\n",
        "        #plt.suptitle(\"Latent Space Interpolation for Glasses, Rosy Cheeks and Goatee\")\n",
        "        #plt.show()\n",
        "        plt.savefig(os.path.join(path, f\"latent_space_interpolation_{np.random.randint(0,999999)}.png\"), dpi=300)\n",
        "        plt.close(fig=fig)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}