digit_transformation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# TODO descriptions..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import datetime\n",
    "\n",
    "import tensorflow as tf\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from utils.image_transformation import random_jitter, normalize, resize\n",
    "from utils.mnist_loader import get_mnist_data_set\n",
    "from utils.model_utils import u_net_generator, discriminator, plot_transformations, UNetLayerConfiguration, \\\n",
    "    DownsampleLayerConfiguration, LayerConfiguration, create_training_function, get_instance_data_set_generator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# settings for training\n",
    "epochs = 300\n",
    "plot_each_nth_epoch = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# create directory for logs\n",
    "logs_base_dir = \"./logs\"\n",
    "os.makedirs(logs_base_dir, exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# create directory for plots\n",
    "output_base_dir = \"./output\"\n",
    "os.makedirs(output_base_dir, exist_ok=True)\n",
    "os.makedirs(output_base_dir+\"/train\", exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# get samples of images with digits '1' and '7' in correct format\n",
    "X_1s, X_7s = get_mnist_data_set()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7ff16c119d68>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# show a sample per digit\n",
    "\n",
    "plt.figure(figsize=(8, 8))\n",
    "\n",
    "plt.subplot(121)\n",
    "plt.title('Sample for digit 1')\n",
    "plt.imshow(normalize(X_1s[0]) * 0.5 + 0.5)\n",
    "\n",
    "plt.subplot(122)\n",
    "plt.title('Sample for digit 7')\n",
    "plt.imshow(normalize(X_7s[0]) * 0.5 + 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7ff15fffe518>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# show a sample per digit with applied random jitter\n",
    "\n",
    "new_image_1 = normalize(random_jitter(X_1s[0], target_size=(32, 32), mirror=False))\n",
    "new_image_7 = normalize(random_jitter(X_7s[0], target_size=(32, 32), mirror=False))\n",
    "\n",
    "plt.figure(figsize=(8, 8))\n",
    "\n",
    "plt.subplot(221)\n",
    "plt.title('Sample for digit 1 with random jitter')\n",
    "plt.imshow(new_image_1 * 0.5 + 0.5)\n",
    "\n",
    "plt.subplot(222)\n",
    "plt.title('Original sample for digit 1')\n",
    "plt.imshow(normalize(X_1s[0]) * 0.5 + 0.5)\n",
    "\n",
    "plt.subplot(223)\n",
    "plt.title('Sample for digit 7 with random jitter')\n",
    "plt.imshow(new_image_7 * 0.5 + 0.5)\n",
    "\n",
    "plt.subplot(224)\n",
    "plt.title('Original sample for digit 7')\n",
    "plt.imshow(normalize(X_7s[0]) * 0.5 + 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# TODO kernel size - gnerator - first and last - increase, discriminator - set to 4\n",
    "\n",
    "# define architecture of networks and initialize networks\n",
    "\n",
    "# how many output channels to have\n",
    "output_channels = 3\n",
    "\n",
    "# define generator architecture\n",
    "u_net_encoder_decoder_layers = tuple([UNetLayerConfiguration(16, 3, apply_norm=False), UNetLayerConfiguration(32, 3), \n",
    "                                      UNetLayerConfiguration(64, 3)])\n",
    "u_net_connecting_layers = tuple([DownsampleLayerConfiguration(64, 3)])\n",
    "\n",
    "# define discriminator architecture\n",
    "discriminator_downsampling_layers = tuple([DownsampleLayerConfiguration(16, 3, apply_norm=False), \n",
    "                                           DownsampleLayerConfiguration(32, 3)])\n",
    "discriminator_final_convolution_layer = LayerConfiguration(64, 3)\n",
    "\n",
    "# initialize generators\n",
    "generator_x2y = u_net_generator(output_channels=output_channels, encoder_decoder_layers=u_net_encoder_decoder_layers,\n",
    "                    connecting_layers=u_net_connecting_layers, input_shape=(32, 32, 3), kernel_size_last_layer=3)\n",
    "generator_y2x = u_net_generator(output_channels=output_channels, encoder_decoder_layers=u_net_encoder_decoder_layers,\n",
    "                    connecting_layers=u_net_connecting_layers, input_shape=(32, 32, 3), kernel_size_last_layer=3)\n",
    "\n",
    "# initialize discriminators\n",
    "discriminator_x = discriminator(input_shape=(32, 32, 3), downsampling_layers=discriminator_downsampling_layers,\n",
    "                                final_convolution_layer=discriminator_final_convolution_layer, kernel_size_last_layer=3)\n",
    "discriminator_y = discriminator(input_shape=(32, 32, 3), downsampling_layers=discriminator_downsampling_layers,\n",
    "                                final_convolution_layer=discriminator_final_convolution_layer, kernel_size_last_layer=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# get data set generators\n",
    "\n",
    "# define function to apply on each data sample\n",
    "function_to_apply_per_train_instance=lambda image: \\\n",
    "    tf.cast([normalize(random_jitter(image, target_size=(32, 32), mirror=False))], tf.float32)\n",
    "function_to_apply_per_test_instance=lambda image: tf.cast([normalize(resize(image, size=(32, 32)))], tf.float32)\n",
    "\n",
    "data_set_generator_x_train = get_instance_data_set_generator(X_1s[:200], function_to_apply_per_train_instance)\n",
    "data_set_generator_x_test = get_instance_data_set_generator(X_1s[200:210], function_to_apply_per_test_instance)\n",
    "data_set_generator_y_train = get_instance_data_set_generator(X_7s[:200], function_to_apply_per_train_instance)\n",
    "data_set_generator_y_test = get_instance_data_set_generator(X_7s[200:210], function_to_apply_per_test_instance)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot transformations (before training)\n",
    "plot_transformations(new_image_1, new_image_7, generator_x2y, generator_y2x, distributions_names=('1s', '7s'), \n",
    "                     fig_size=(16, 10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# init. summary writer\n",
    "summary_writer = tf.summary.create_file_writer(logs_base_dir + \"/\" + datetime.datetime.now().strftime(\"%Y%m%d-%H%M%S\"))\n",
    "\n",
    "# init. training epoch executor\n",
    "training_epoch_executor = create_training_function(data_set_generator_x_train, data_set_generator_y_train, 200, \n",
    "                                                   generator_x2y, discriminator_x, generator_y2x, discriminator_y, \n",
    "                                                   summary_writer=summary_writer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# run training for N number of epochs\n",
    "\n",
    "sum_time = 0\n",
    "\n",
    "for i in range(epochs):\n",
    "    epoch, tm = next(training_epoch_executor)\n",
    "    sum_time = sum_time + tm\n",
    "    if (epoch + 1) % plot_each_nth_epoch == 0:\n",
    "        # plot the most recent results\n",
    "        plot_title = \"Generated data in iteration {}. Average time per epoch was {}s.\"\\\n",
    "            .format(epoch+1, sum_time / plot_each_nth_epoch)\n",
    "        save_as = \"{}/{}/{}_epoch.jpg\".format(output_base_dir, \"train\", epoch+1)\n",
    "        plot_transformations(next(data_set_generator_x_test)[0], next(data_set_generator_y_test)[0], generator_x2y, \n",
    "                             generator_y2x, distributions_names=('1s', '7s'), main_title=plot_title, \n",
    "                             save_as=save_as, fig_size=(16, 10))\n",
    "        # reset timer\n",
    "        sum_time = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# plot transformations (after training)\n",
    "\n",
    "for i in range(10):\n",
    "    x_instance = next(data_set_generator_x_test)[0]\n",
    "    y_instance = next(data_set_generator_y_test)[0]\n",
    "    save_as = \"{}/{}_instance.jpg\".format(output_base_dir, i)\n",
    "    plot_transformations(x_instance, y_instance, generator_x2y, generator_y2x, distributions_names=('1s', '7s'), \n",
    "                         fig_size=(16, 10), save_as=save_as)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "metadata": {
     "collapsed": false
    },
    "source": []
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}