{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "%matplotlib inline"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "\n# Pottok for image color adaptation with labels - RasterOptimalTransport\n\nUsing sinkhorn L1l2\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "import numpy as np\nimport matplotlib.pylab as pl\nimport ot\nimport pottok\nfrom sklearn.preprocessing import StandardScaler,MinMaxScaler  # centrer-r\u00e9duire\n\n\n\nsource_image,source_vector,target_image,target_vector = pottok.datasets.load_pottoks(return_only_path = True)\n\nbrown_pottok,black_pottok = pottok.datasets.load_pottoks(return_X_y=False)\nbrown_pottok = brown_pottok/255\nblack_pottok = black_pottok/255\n\n\nlabel = 'level'"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Optimal transport with SinkhornL1l2 with circular gridsearch\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "raster_transport_circular = pottok.RasterOptimalTransport(transport_function=ot.da.SinkhornL1l2Transport,\n                                        params=dict(reg_e=[1e-1,1e-0], reg_cl=[1e-1]))\nraster_transport_circular.preprocessing(image_source = source_image,\n                   image_target = target_image,\n                   vector_source = source_vector,\n                   vector_target = target_vector,\n                   label_source = label,\n                   label_target = label,\n                   scaler = MinMaxScaler)\n\n\nraster_transport_circular.fit_circular()\n\n\n# Best grid is {'reg_e': 1.0, 'reg_cl': 1.0}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Plot images\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "Xt_transp_unscaled,  Xt_transp_scaled = raster_transport_circular.predict_transfer(raster_transport_circular.source)\n\npl.figure(1, figsize=(10,8))\n\npl.subplot(2, 2, 1)\npl.imshow(brown_pottok)\npl.axis('off')\npl.title('Brown pottok (Source)')\n\npl.subplot(2, 2, 3)\npl.imshow(black_pottok)\npl.axis('off')\npl.title('Black pottok (Target)')\n\npl.subplot(2, 2, 4)\npl.imshow(Xt_transp_unscaled.reshape(*brown_pottok.shape)/255)\npl.axis('off')\npl.title('SinkhornL1l2 (Source to Target with labels)')\n\npl.show()\n\n\n\n\n# ##############################################################################\n# # Optimal transport with SinkhornL1l2 with crossed gridsearch\n# # --------------------------------------------------------------\n\nraster_transport_crossed = pottok.RasterOptimalTransport(transport_function=ot.da.SinkhornL1l2Transport,\n                                        params=dict(reg_e=[1e-1,1e-0], reg_cl=[1e-1]))\nraster_transport_crossed.preprocessing(image_source = source_image,\n                   image_target = target_image,\n                   vector_source = source_vector,\n                   vector_target = target_vector,\n                   label_source = label,\n                   label_target = label,\n                   scaler = MinMaxScaler)\n\n\nraster_transport_crossed.fit_crossed()\n\n\n# Best grid is {'reg_e': 0.1, 'reg_cl': 0.1}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## Plot images\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "Xt_transp_unscaled,  Xt_transp_scaled = raster_transport_crossed.predict_transfer(raster_transport_crossed.source)\n\npl.figure(2, figsize=(10,8))\n\npl.subplot(2, 2, 1)\npl.imshow(brown_pottok)\npl.axis('off')\npl.title('Brown pottok (Source)')\n\npl.subplot(2, 2, 3)\npl.imshow(black_pottok)\npl.axis('off')\npl.title('Black pottok (Target)')\n\npl.subplot(2, 2, 4)\npl.imshow(Xt_transp_scaled.reshape(*brown_pottok.shape))\npl.axis('off')\npl.title('SinkhornL1l2 (Source to Target with labels)')\n\npl.show()"
      ]
    }
  ],
  "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": "3.7.7"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}