notebooks/Mean test results plot.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "629e9da0",
   "metadata": {},
   "source": [
    "# Mean results plot\n",
    "Simple notebook to plot figures for the paper.\n",
    "\n",
    "Authors:\n",
    "Edoardo Daniele Cannas - edoardodaniele.cannas@polimi.it  \n",
    "Sriram Baireddy - sbairedd@purdue.edu  \n",
    "Paolo Bestagini - paolo.bestagini@polimi.it  \n",
    "Stefano Tubaro - stefano.tubaro@polimi.it  \n",
    "Edward J. Delp - ace@purdue.edu  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdbd0256",
   "metadata": {},
   "source": [
    "## Libraries import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "681591e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "import glob\n",
    "import cv2\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from matplotlib import rc\n",
    "rc('font',**{'family':'sans-serif','sans-serif':['Times New Roman']})\n",
    "## for Palatino and other serif fonts use:\n",
    "#rc('font',**{'family':'serif','serif':['Palatino']})\n",
    "rc('text', usetex=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3ccbdaa2",
   "metadata": {},
   "source": [
    "## Helpers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "56465fd4",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import RobustScaler, QuantileTransformer\n",
    "\n",
    "def quantile_scaling(img: np.ndarray, dist: str) -> np.ndarray:\n",
    "\t\"\"\"\n",
    "\tQuantileTransformer: maps the image range to a specific stastical distribution\n",
    "\t\n",
    "\t:param img: np.ndarray, input image (tested GRD 16-bits)\n",
    "\t:param distribution: str, distribution for mapping the image (either 'uniform' or 'normal')\n",
    "\t:return: np.ndarray, normalized numpy array\n",
    "\t\"\"\"\n",
    "\treturn QuantileTransformer(output_distribution=dist).fit_transform(img.reshape(-1, 1)).reshape(img.shape)\n",
    "\n",
    "\n",
    "def rename_sat(product: str) -> str:\n",
    "    if product[0] == 'L':\n",
    "        sat_name = 'Landsat8'\n",
    "    else:\n",
    "        sat_name = product.split('_')[0]\n",
    "        if sat_name in ['AMAZONIA', 'CBERS']:\n",
    "            sat_name += f'{product.split(\"_\")[1]}_{product.split(\"_\")[2]}'\n",
    "    return sat_name\n",
    "\n",
    "\n",
    "def compute_model_sign(train_tag: str) -> str:\n",
    "    # Create a tag for the visualization\n",
    "    model_sign = train_tag.split('/')[-3]\n",
    "    return model_sign\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "49e4b906",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "(function(on) {\n",
       "const e=$( \"<a>Setup failed</a>\" );\n",
       "const ns=\"js_jupyter_suppress_warnings\";\n",
       "var cssrules=$(\"#\"+ns);\n",
       "if(!cssrules.length) cssrules = $(\"<style id='\"+ns+\"' type='text/css'>div.output_stderr { } </style>\").appendTo(\"head\");\n",
       "e.click(function() {\n",
       "    var s='Showing';  \n",
       "    cssrules.empty()\n",
       "    if(on) {\n",
       "        s='Hiding';\n",
       "        cssrules.append(\"div.output_stderr, div[data-mime-type*='.stderr'] { display:none; }\");\n",
       "    }\n",
       "    e.text(s+' warnings (click to toggle)');\n",
       "    on=!on;\n",
       "}).click();\n",
       "$(element).append(e);\n",
       "})(true);\n"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%javascript\n",
    "(function(on) {\n",
    "const e=$( \"<a>Setup failed</a>\" );\n",
    "const ns=\"js_jupyter_suppress_warnings\";\n",
    "var cssrules=$(\"#\"+ns);\n",
    "if(!cssrules.length) cssrules = $(\"<style id='\"+ns+\"' type='text/css'>div.output_stderr { } </style>\").appendTo(\"head\");\n",
    "e.click(function() {\n",
    "    var s='Showing';  \n",
    "    cssrules.empty()\n",
    "    if(on) {\n",
    "        s='Hiding';\n",
    "        cssrules.append(\"div.output_stderr, div[data-mime-type*='.stderr'] { display:none; }\");\n",
    "    }\n",
    "    e.text(s+' warnings (click to toggle)');\n",
    "    on=!on;\n",
    "}).click();\n",
    "$(element).append(e);\n",
    "})(true);"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b165758a069acf33",
   "metadata": {},
   "source": [
    "## Load the results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "784ca26c",
   "metadata": {},
   "source": [
    "### Histogram Equalization Generated Dataset (HEGD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e89145c6",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_451295/1282501006.py:20: FutureWarning: The default value of numeric_only in DataFrameGroupBy.mean is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
      "  he_mean_results = all_results_df.groupby(['Model_name', 'Operation']).mean()[['AUC']]\n"
     ]
    }
   ],
   "source": [
    "# Look for all models used\n",
    "results_dir = '../results/hegd'  # PLACE YOUR RESULTS DIRECTORY HERE!\n",
    "all_dfs_paths = glob.glob(os.path.join(results_dir, '**', 'results_df_mean_fingerprints.pkl'), recursive=True)\n",
    "\n",
    "# Load the Dataframe for each model and put it in a bigger one\n",
    "all_results_df = []\n",
    "for path in all_dfs_paths:\n",
    "    # Load results\n",
    "    df = pd.read_pickle(path)\n",
    "    # Create a tag for the visualization\n",
    "    model_sign = compute_model_sign(path)\n",
    "    # Save in the bigger DataFrame\n",
    "    df = pd.concat({model_sign: df}, names=['Model_name', 'Operation', 'Img_idx'])\n",
    "    all_results_df.append(df)\n",
    "\n",
    "# Concatenate everything\n",
    "all_results_df = pd.concat(all_results_df)\n",
    "\n",
    "# Compute the mean across operations\n",
    "he_mean_results = all_results_df.groupby(['Model_name', 'Operation']).mean()[['AUC']]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "67605bd9",
   "metadata": {},
   "source": [
    "### Standard Generated Dataset (SGD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b34880da",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_451295/3642307309.py:20: FutureWarning: The default value of numeric_only in DataFrameGroupBy.mean is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
      "  max_abs_mean_results = all_results_df.groupby(['Model_name', 'Operation']).mean()[['AUC']]\n"
     ]
    }
   ],
   "source": [
    "# Look for all models used\n",
    "results_dir = '../results/sgd'  # PLACE YOUR RESULTS DIRECTORY HERE!\n",
    "all_dfs_paths = glob.glob(os.path.join(results_dir, '**', 'results_df_mean_fingerprints.pkl'), recursive=True)\n",
    "\n",
    "# Load the Dataframe for each model and put it in a bigger one\n",
    "all_results_df = []\n",
    "for path in all_dfs_paths:\n",
    "    # Load results\n",
    "    df = pd.read_pickle(path)\n",
    "    # Create a tag for the visualization\n",
    "    model_sign = compute_model_sign(path)\n",
    "    # Save in the bigger DataFrame\n",
    "    df = pd.concat({model_sign: df}, names=['Model_name', 'Operation', 'Img_idx'])\n",
    "    all_results_df.append(df)\n",
    "\n",
    "# Concatenate everything\n",
    "all_results_df = pd.concat(all_results_df)\n",
    "\n",
    "# Compute the mean across operations\n",
    "max_abs_mean_results = all_results_df.groupby(['Model_name', 'Operation']).mean()[['AUC']]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b56830fd",
   "metadata": {},
   "source": [
    "### Join together the Dataframes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "33959482",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>AUC SG</th>\n",
       "      <th>AUC HEG</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Model_name</th>\n",
       "      <th>Operation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">HistogramEqualization</th>\n",
       "      <th>UnnamedAffine</th>\n",
       "      <td>0.756331</td>\n",
       "      <td>0.763947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedAverageBlur</th>\n",
       "      <td>0.744204</td>\n",
       "      <td>0.742003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedGaussianBlur</th>\n",
       "      <td>0.758412</td>\n",
       "      <td>0.750998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedIdentity</th>\n",
       "      <td>0.760556</td>\n",
       "      <td>0.783793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedLogContrast</th>\n",
       "      <td>0.772095</td>\n",
       "      <td>0.768795</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedMotionBlur</th>\n",
       "      <td>0.757253</td>\n",
       "      <td>0.769163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPerspectiveTransform</th>\n",
       "      <td>0.761552</td>\n",
       "      <td>0.764107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPiecewiseAffine</th>\n",
       "      <td>0.763977</td>\n",
       "      <td>0.766819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedSigmoidContrast</th>\n",
       "      <td>0.761040</td>\n",
       "      <td>0.777605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MaxScaling</th>\n",
       "      <th>UnnamedAffine</th>\n",
       "      <td>0.742213</td>\n",
       "      <td>0.752721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedAverageBlur</th>\n",
       "      <td>0.757626</td>\n",
       "      <td>0.728235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedGaussianBlur</th>\n",
       "      <td>0.749822</td>\n",
       "      <td>0.741118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedIdentity</th>\n",
       "      <td>0.770409</td>\n",
       "      <td>0.771507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedLogContrast</th>\n",
       "      <td>0.763800</td>\n",
       "      <td>0.771489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedMotionBlur</th>\n",
       "      <td>0.756712</td>\n",
       "      <td>0.738521</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPerspectiveTransform</th>\n",
       "      <td>0.750572</td>\n",
       "      <td>0.750220</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPiecewiseAffine</th>\n",
       "      <td>0.742307</td>\n",
       "      <td>0.752633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedSigmoidContrast</th>\n",
       "      <td>0.772171</td>\n",
       "      <td>0.772049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MinPMax_95</th>\n",
       "      <th>UnnamedAffine</th>\n",
       "      <td>0.878123</td>\n",
       "      <td>0.726338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedAverageBlur</th>\n",
       "      <td>0.894674</td>\n",
       "      <td>0.726864</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedGaussianBlur</th>\n",
       "      <td>0.890287</td>\n",
       "      <td>0.729464</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedIdentity</th>\n",
       "      <td>0.883864</td>\n",
       "      <td>0.725904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedLogContrast</th>\n",
       "      <td>0.903303</td>\n",
       "      <td>0.732466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedMotionBlur</th>\n",
       "      <td>0.883550</td>\n",
       "      <td>0.734275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPerspectiveTransform</th>\n",
       "      <td>0.885665</td>\n",
       "      <td>0.732015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPiecewiseAffine</th>\n",
       "      <td>0.890646</td>\n",
       "      <td>0.727450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedSigmoidContrast</th>\n",
       "      <td>0.925032</td>\n",
       "      <td>0.754774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MinPMax_99</th>\n",
       "      <th>UnnamedAffine</th>\n",
       "      <td>0.858914</td>\n",
       "      <td>0.636660</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedAverageBlur</th>\n",
       "      <td>0.879219</td>\n",
       "      <td>0.638835</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedGaussianBlur</th>\n",
       "      <td>0.872064</td>\n",
       "      <td>0.637977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedIdentity</th>\n",
       "      <td>0.864366</td>\n",
       "      <td>0.627199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedLogContrast</th>\n",
       "      <td>0.871211</td>\n",
       "      <td>0.649995</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedMotionBlur</th>\n",
       "      <td>0.869927</td>\n",
       "      <td>0.627036</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPerspectiveTransform</th>\n",
       "      <td>0.872449</td>\n",
       "      <td>0.632484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedPiecewiseAffine</th>\n",
       "      <td>0.878822</td>\n",
       "      <td>0.631335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UnnamedSigmoidContrast</th>\n",
       "      <td>0.896265</td>\n",
       "      <td>0.660360</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                     AUC SG   AUC HEG\n",
       "Model_name            Operation                                      \n",
       "HistogramEqualization UnnamedAffine                0.756331  0.763947\n",
       "                      UnnamedAverageBlur           0.744204  0.742003\n",
       "                      UnnamedGaussianBlur          0.758412  0.750998\n",
       "                      UnnamedIdentity              0.760556  0.783793\n",
       "                      UnnamedLogContrast           0.772095  0.768795\n",
       "                      UnnamedMotionBlur            0.757253  0.769163\n",
       "                      UnnamedPerspectiveTransform  0.761552  0.764107\n",
       "                      UnnamedPiecewiseAffine       0.763977  0.766819\n",
       "                      UnnamedSigmoidContrast       0.761040  0.777605\n",
       "MaxScaling            UnnamedAffine                0.742213  0.752721\n",
       "                      UnnamedAverageBlur           0.757626  0.728235\n",
       "                      UnnamedGaussianBlur          0.749822  0.741118\n",
       "                      UnnamedIdentity              0.770409  0.771507\n",
       "                      UnnamedLogContrast           0.763800  0.771489\n",
       "                      UnnamedMotionBlur            0.756712  0.738521\n",
       "                      UnnamedPerspectiveTransform  0.750572  0.750220\n",
       "                      UnnamedPiecewiseAffine       0.742307  0.752633\n",
       "                      UnnamedSigmoidContrast       0.772171  0.772049\n",
       "MinPMax_95            UnnamedAffine                0.878123  0.726338\n",
       "                      UnnamedAverageBlur           0.894674  0.726864\n",
       "                      UnnamedGaussianBlur          0.890287  0.729464\n",
       "                      UnnamedIdentity              0.883864  0.725904\n",
       "                      UnnamedLogContrast           0.903303  0.732466\n",
       "                      UnnamedMotionBlur            0.883550  0.734275\n",
       "                      UnnamedPerspectiveTransform  0.885665  0.732015\n",
       "                      UnnamedPiecewiseAffine       0.890646  0.727450\n",
       "                      UnnamedSigmoidContrast       0.925032  0.754774\n",
       "MinPMax_99            UnnamedAffine                0.858914  0.636660\n",
       "                      UnnamedAverageBlur           0.879219  0.638835\n",
       "                      UnnamedGaussianBlur          0.872064  0.637977\n",
       "                      UnnamedIdentity              0.864366  0.627199\n",
       "                      UnnamedLogContrast           0.871211  0.649995\n",
       "                      UnnamedMotionBlur            0.869927  0.627036\n",
       "                      UnnamedPerspectiveTransform  0.872449  0.632484\n",
       "                      UnnamedPiecewiseAffine       0.878822  0.631335\n",
       "                      UnnamedSigmoidContrast       0.896265  0.660360"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_results = max_abs_mean_results.join(he_mean_results, rsuffix=' HEG', lsuffix=' SG')\n",
    "all_results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "89988d92",
   "metadata": {},
   "source": [
    "## Plot the results\n",
    "We are going to have a separate bar plot for the SGD and the HEGD datasets."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "61631c15f43e167a",
   "metadata": {},
   "source": [
    "### Re-organize the Dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "11a26319",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>AUC SG</th>\n",
       "      <th>AUC HEG</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Preprocessing strategy</th>\n",
       "      <th>Operation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">HistogramEqualization</th>\n",
       "      <th>Average blur</th>\n",
       "      <td>0.744204</td>\n",
       "      <td>0.742003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Gaussian Blur</th>\n",
       "      <td>0.758412</td>\n",
       "      <td>0.750998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Log contrast</th>\n",
       "      <td>0.772095</td>\n",
       "      <td>0.768795</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Motion blur</th>\n",
       "      <td>0.757253</td>\n",
       "      <td>0.769163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No editing</th>\n",
       "      <td>0.760556</td>\n",
       "      <td>0.783793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Piecewise affine</th>\n",
       "      <td>0.763977</td>\n",
       "      <td>0.766819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prospective affine</th>\n",
       "      <td>0.761552</td>\n",
       "      <td>0.764107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rotation + resize</th>\n",
       "      <td>0.756331</td>\n",
       "      <td>0.763947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sigmoid contrast</th>\n",
       "      <td>0.761040</td>\n",
       "      <td>0.777605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MaxScaling</th>\n",
       "      <th>Average blur</th>\n",
       "      <td>0.757626</td>\n",
       "      <td>0.728235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Gaussian Blur</th>\n",
       "      <td>0.749822</td>\n",
       "      <td>0.741118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Log contrast</th>\n",
       "      <td>0.763800</td>\n",
       "      <td>0.771489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Motion blur</th>\n",
       "      <td>0.756712</td>\n",
       "      <td>0.738521</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No editing</th>\n",
       "      <td>0.770409</td>\n",
       "      <td>0.771507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Piecewise affine</th>\n",
       "      <td>0.742307</td>\n",
       "      <td>0.752633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prospective affine</th>\n",
       "      <td>0.750572</td>\n",
       "      <td>0.750220</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rotation + resize</th>\n",
       "      <td>0.742213</td>\n",
       "      <td>0.752721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sigmoid contrast</th>\n",
       "      <td>0.772171</td>\n",
       "      <td>0.772049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MinPMax $95^{th}$ percentile</th>\n",
       "      <th>Average blur</th>\n",
       "      <td>0.894674</td>\n",
       "      <td>0.726864</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Gaussian Blur</th>\n",
       "      <td>0.890287</td>\n",
       "      <td>0.729464</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Log contrast</th>\n",
       "      <td>0.903303</td>\n",
       "      <td>0.732466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Motion blur</th>\n",
       "      <td>0.883550</td>\n",
       "      <td>0.734275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No editing</th>\n",
       "      <td>0.883864</td>\n",
       "      <td>0.725904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Piecewise affine</th>\n",
       "      <td>0.890646</td>\n",
       "      <td>0.727450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prospective affine</th>\n",
       "      <td>0.885665</td>\n",
       "      <td>0.732015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rotation + resize</th>\n",
       "      <td>0.878123</td>\n",
       "      <td>0.726338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sigmoid contrast</th>\n",
       "      <td>0.925032</td>\n",
       "      <td>0.754774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"9\" valign=\"top\">MinPMax $99^{th}$ percentile</th>\n",
       "      <th>Average blur</th>\n",
       "      <td>0.879219</td>\n",
       "      <td>0.638835</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Gaussian Blur</th>\n",
       "      <td>0.872064</td>\n",
       "      <td>0.637977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Log contrast</th>\n",
       "      <td>0.871211</td>\n",
       "      <td>0.649995</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Motion blur</th>\n",
       "      <td>0.869927</td>\n",
       "      <td>0.627036</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No editing</th>\n",
       "      <td>0.864366</td>\n",
       "      <td>0.627199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Piecewise affine</th>\n",
       "      <td>0.878822</td>\n",
       "      <td>0.631335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prospective affine</th>\n",
       "      <td>0.872449</td>\n",
       "      <td>0.632484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rotation + resize</th>\n",
       "      <td>0.858914</td>\n",
       "      <td>0.636660</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sigmoid contrast</th>\n",
       "      <td>0.896265</td>\n",
       "      <td>0.660360</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                   AUC SG   AUC HEG\n",
       "Preprocessing strategy       Operation                             \n",
       "HistogramEqualization        Average blur        0.744204  0.742003\n",
       "                             Gaussian Blur       0.758412  0.750998\n",
       "                             Log contrast        0.772095  0.768795\n",
       "                             Motion blur         0.757253  0.769163\n",
       "                             No editing          0.760556  0.783793\n",
       "                             Piecewise affine    0.763977  0.766819\n",
       "                             Prospective affine  0.761552  0.764107\n",
       "                             Rotation + resize   0.756331  0.763947\n",
       "                             Sigmoid contrast    0.761040  0.777605\n",
       "MaxScaling                   Average blur        0.757626  0.728235\n",
       "                             Gaussian Blur       0.749822  0.741118\n",
       "                             Log contrast        0.763800  0.771489\n",
       "                             Motion blur         0.756712  0.738521\n",
       "                             No editing          0.770409  0.771507\n",
       "                             Piecewise affine    0.742307  0.752633\n",
       "                             Prospective affine  0.750572  0.750220\n",
       "                             Rotation + resize   0.742213  0.752721\n",
       "                             Sigmoid contrast    0.772171  0.772049\n",
       "MinPMax $95^{th}$ percentile Average blur        0.894674  0.726864\n",
       "                             Gaussian Blur       0.890287  0.729464\n",
       "                             Log contrast        0.903303  0.732466\n",
       "                             Motion blur         0.883550  0.734275\n",
       "                             No editing          0.883864  0.725904\n",
       "                             Piecewise affine    0.890646  0.727450\n",
       "                             Prospective affine  0.885665  0.732015\n",
       "                             Rotation + resize   0.878123  0.726338\n",
       "                             Sigmoid contrast    0.925032  0.754774\n",
       "MinPMax $99^{th}$ percentile Average blur        0.879219  0.638835\n",
       "                             Gaussian Blur       0.872064  0.637977\n",
       "                             Log contrast        0.871211  0.649995\n",
       "                             Motion blur         0.869927  0.627036\n",
       "                             No editing          0.864366  0.627199\n",
       "                             Piecewise affine    0.878822  0.631335\n",
       "                             Prospective affine  0.872449  0.632484\n",
       "                             Rotation + resize   0.858914  0.636660\n",
       "                             Sigmoid contrast    0.896265  0.660360"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select the models to plot\n",
    "plot_df = all_results.loc[['MinPMax_95', 'MinPMax_99', 'MaxScaling', 'HistogramEqualization']]\n",
    "# Rename the columns\n",
    "plot_df = plot_df.rename({'MinPMax_95': r'MinPMax $95^{th}$ percentile',\n",
    "                          'MinPMax_99': r'MinPMax $99^{th}$ percentile',\n",
    "                          'MaxScaling': r'MaxScaling',\n",
    "                          'HistogramEqualization': 'HistogramEqualization'})\n",
    "# Rename the index\n",
    "plot_df = plot_df.rename_axis(index={'Model_name': 'Preprocessing strategy'})\n",
    "# Rename the operations\n",
    "operations_mapping = {'Affine': 'Rotation + resize', 'AverageBlur': 'Average blur', 'GaussianBlur': 'Gaussian Blur',\n",
    "                      'Identity': 'No editing', 'LogContrast': 'Log contrast', 'MotionBlur': 'Motion blur',\n",
    "                      'PerspectiveTransform': 'Prospective affine', 'PiecewiseAffine': 'Piecewise affine', \n",
    "                      'SigmoidContrast': 'Sigmoid contrast'}\n",
    "plot_df.index = plot_df.index.set_levels([plot_df.index.levels[0], plot_df.index.levels[1].map(lambda x: operations_mapping[x.replace('Unnamed', '')])])\n",
    "plot_df = plot_df.sort_index()\n",
    "plot_df\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "653e44eaf0aac258",
   "metadata": {},
   "source": [
    "### Plot the SGD results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "cb7f519e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 700x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot params\n",
    "plt.rcParams['font.family'] = 'Times New Roman'\n",
    "fontsize=20\n",
    "ax = plot_df['AUC SG'].swaplevel(0, 1).unstack(level=1).plot.barh(legend=True, \n",
    "                                                                 xlim=[0, 0.95], \n",
    "                                                                 xticks=np.linspace(0, 1.0, 11), \n",
    "                                                                 grid=True, figsize=(7, 5))\n",
    "\n",
    "# Adjust the layout\n",
    "ax.tick_params(axis='both', labelsize=fontsize-5)\n",
    "ax.set_ylabel('Editing operation', fontsize=fontsize)\n",
    "ax.set_xlabel('AUC value', fontsize=fontsize)\n",
    "ax.axvline(x=plot_df['AUC SG'].mean(), color='black', linestyle='--', label='Mean performance')\n",
    "legend = ax.legend(loc='upper left', prop={'size': fontsize-5})\n",
    "legend.get_frame().set_facecolor('white')\n",
    "legend.get_frame().set_alpha(1)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93cf40e8aa5a5743",
   "metadata": {},
   "source": [
    "### Plot the HEGD results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4f3508cf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 700x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot params\n",
    "plt.rcParams['font.family'] = 'Times New Roman'\n",
    "fontsize=20\n",
    "ax = plot_df['AUC HEG'].swaplevel(0, 1).unstack(level=1).plot.barh(legend=True, \n",
    "                                                                 xlim=[0, 0.95], \n",
    "                                                                 xticks=np.linspace(0, 1.0, 11), \n",
    "                                                                 grid=True, figsize=(7, 5))\n",
    "\n",
    "# Adjust the layout\n",
    "ax.tick_params(axis='both', labelsize=fontsize-5)\n",
    "ax.set_ylabel('Editing operation', fontsize=fontsize)\n",
    "ax.set_xlabel('AUC value', fontsize=fontsize)\n",
    "ax.axvline(x=plot_df['AUC HEG'].mean(), color='black', linestyle='--', label='Mean performance')\n",
    "legend = ax.legend(loc='upper left', prop={'size': fontsize-5})\n",
    "legend.get_frame().set_facecolor('white')\n",
    "legend.get_frame().set_alpha(1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7a699329",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}