ld2daps/Notebooks/CrossVal spatial.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sys.path.append('../triskele/python/')\n",
    "import triskele\n",
    "\n",
    "figsize = np.array((16, 3)) * 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Spatial Cross Validation Generator\n",
    "\n",
    "Due to the operating of attribute profiles, we need a spatial cross validation generator to ensure we do not train and test samples with the same root in the hierarchical representation used during the description process."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load ground truth & metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "gt = triskele.read('../Data/ground_truth/2018_IEEE_GRSS_DFC_GT_TR.tif')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=figsize)\n",
    "plt.imshow(gt, cmap=plt.get_cmap('GnBu'))\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfc_lbl  = pd.read_csv('../GroundTruth/labels.csv')\n",
    "meta_idx = pd.read_csv('../GroundTruth/jurse_meta_idx.csv')\n",
    "meta_lbl = pd.read_csv('../GroundTruth/jurse_meta_lbl.csv')\n",
    "\n",
    "imap = np.array(meta_idx['metaclass_index'])\n",
    "\n",
    "all_meta_view = dfc_lbl.merge(meta_idx).merge(meta_lbl)\n",
    "all_meta_view"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=figsize)\n",
    "plt.imshow(imap[gt], cmap=plt.get_cmap('GnBu'))\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Labels map"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axs = plt.subplots(dfc_lbl['label'].size, figsize=figsize * dfc_lbl['label'].size)\n",
    "\n",
    "for i, lbl in enumerate(dfc_lbl['label']):\n",
    "    omap = gt == np.array(dfc_lbl[dfc_lbl['label'] == lbl ]['index'])\n",
    "    axs[i].set_title(lbl)\n",
    "    axs[i].imshow(1. * omap, cmap=plt.get_cmap('GnBu'))\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Meta labels map"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axs = plt.subplots(meta_lbl['metaclass_label'].size, figsize=figsize * meta_lbl['metaclass_label'].size)\n",
    "\n",
    "for i, lbl in enumerate(meta_lbl['metaclass_label']):\n",
    "    omap = np.isin(gt, np.array(all_meta_view[all_meta_view['metaclass_label'] == lbl]['index']))\n",
    "    axs[i].set_title(lbl)\n",
    "    axs[i].imshow(1. * omap, cmap=plt.get_cmap('GnBu'))\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "col, count = np.unique(cut, return_counts=True)\n",
    "df = pd.DataFrame(count[np.newaxis,:], columns=col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "count = 4\n",
    "\n",
    "step = int(gt.shape[0] / count)\n",
    "\n",
    "fig, axs = plt.subplots(count, figsize=figsize * count)\n",
    "\n",
    "col, count_all = np.unique(imap[gt], return_counts=True)\n",
    "\n",
    "metrics = pd.DataFrame(columns=col)\n",
    "counts = list()\n",
    "for i in range(count):\n",
    "    cut = imap[gt][i*step:(i+1)*step+1]\n",
    "    axs[i].imshow(cut, cmap=plt.get_cmap('GnBu'))\n",
    "    axs[i].set_title('Cut {}'.format(i))\n",
    "    \n",
    "    col, count = np.unique(cut, return_counts=True)\n",
    "    df = pd.DataFrame(count[np.newaxis,:], columns=col, index=[i])\n",
    "    counts.append(df)\n",
    "\n",
    "plt.savefig('../Res/meta.png',bbox_inches='tight', transparent=\"True\", pad_inches=0)\n",
    "plt.show()\n",
    "\n",
    "metrics = pd.concat(counts)\n",
    "metrics / count_all * 100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Try to remove filtered object"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import ld2dap"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tif = ld2dap.LoadTIFF('../Data/phase1_rasters/DSM_C12/UH17c_GEF051_TR.tif')\n",
    "trh = ld2dap.Treshold(70)\n",
    "dsp = ld2dap.ShowFig(symb=True)\n",
    "\n",
    "aps = ld2dap.SelfDualAttributeProfiles(area=[100, 1000, 1e4, 1e5, 1e6, 1e7])\n",
    "dif = ld2dap.Differential()\n",
    "\n",
    "dif.input = aps\n",
    "aps.input = trh\n",
    "dsp.input = dif\n",
    "trh.input = tif\n",
    "\n",
    "dsp.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tif = ld2dap.LoadTIFF('../Data/phase1_rasters/DSM_C12/UH17c_GEF051_TR.tif')\n",
    "#trh = ld2dap.Treshold(1e4)\n",
    "aps = ld2dap.SelfDualAttributeProfiles(area=[100, 1e3, 1e4])\n",
    "dsp = ld2dap.ShowFig(symb=True)\n",
    "dif = ld2dap.Differential()\n",
    "ddsp = ld2dap.ShowFig(symb=True)\n",
    "\n",
    "ddsp.input = dif\n",
    "dif.input = aps\n",
    "dsp.input = aps\n",
    "aps.input = tif\n",
    "#trh.input = tif\n",
    "\n",
    "tif.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}