preprocess_yolo_labels_3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "v2. This time output center point of ROI instead of upper left corner.\n",
    "v3. Center x, y, w and h values around mean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as patches\n",
    "import pandas as pd\n",
    "import csv\n",
    "import pickle\n",
    "import os\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "GRID_SIZE = 4\n",
    "location_data = np.zeros((GRID_SIZE,GRID_SIZE,5))\n",
    "\n",
    "for i in range(4):\n",
    "    for j in range(4):\n",
    "        location_data[i,j,:] = [0, 390, 365, 225, 335]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "coords = np.arange(0, 1024, 256)\n",
    "\n",
    "fig,ax = plt.subplots(1, figsize=(10,10))\n",
    "\n",
    "rect_1 = patches.Rectangle((195,511),218,271,linewidth=1,edgecolor='r',facecolor='none')\n",
    "plt.plot(304, 646.5, marker=\"o\", color=\"r\")\n",
    "rect_2 = patches.Rectangle((612,214),109,89,linewidth=1,edgecolor='b',facecolor='none')\n",
    "plt.plot(666.5, 258.5, marker=\"o\", color=\"b\")\n",
    "rect_3 = patches.Rectangle((610,599),265,245,linewidth=1,edgecolor='g',facecolor='none')\n",
    "plt.plot(742.5, 721.5, marker=\"o\", color=\"g\")\n",
    "ax.add_patch(rect_1)\n",
    "ax.add_patch(rect_2)\n",
    "ax.add_patch(rect_3)\n",
    "plt.xticks(coords)\n",
    "plt.yticks(coords)\n",
    "for item in coords:\n",
    "    plt.axvline(item, linewidth=0.5)\n",
    "    plt.axhline(item, linewidth=0.5)\n",
    "plt.xlim(0,1024)\n",
    "plt.ylim(0,1024)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(4.0, 10.0), (10.0, 4.0), (11.0, 11.0)]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "boxes = [\n",
    "    [195,511,218,271],\n",
    "    [612,214,109,89],\n",
    "    [610,599,265,245]\n",
    "]\n",
    "\n",
    "centers = []\n",
    "\n",
    "for box in boxes:\n",
    "    x,y,w,h = box\n",
    "    center_x = x + (w / 2)\n",
    "    center_y = y + (h / 2)\n",
    "    centers.append((center_x,center_y))\n",
    "\n",
    "pixel_locs = []    \n",
    "for center in centers:\n",
    "    x_pixel = center[0] // 64\n",
    "    y_pixel = center[1] // 64\n",
    "    \n",
    "    pixel_locs.append((x_pixel,y_pixel))\n",
    "    \n",
    "pixel_locs    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train dir: ./stage_1_train_images\n"
     ]
    }
   ],
   "source": [
    "ROOT_DIR = \"./\"\n",
    "\n",
    "train_dicom_dir = os.path.join(ROOT_DIR, 'stage_1_train_images')\n",
    "test_dicom_dir = os.path.join(ROOT_DIR, 'stage_1_test_images')\n",
    "print(\"Train dir:\", train_dicom_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "GRID_SIZE = 4\n",
    "CELL_WIDTH = 1024 // GRID_SIZE\n",
    "\n",
    "# means to center data\n",
    "mu_x = 391.456158\n",
    "mu_y = 363.1358768\n",
    "mu_w = 220.8453815\n",
    "mu_h = 334.1743641\n",
    "mu_center_x = 501.8788487\n",
    "mu_center_y = 530.2230589\n",
    "\n",
    "\n",
    "# empty dictionary\n",
    "pneumonia_locations = {}\n",
    "used_filenames = []\n",
    "pixel_means = []\n",
    "locations = []\n",
    "\n",
    "# load table\n",
    "with open(os.path.join('./stage_1_train_labels.csv'), mode='r') as infile:\n",
    "    # open reader\n",
    "    reader = csv.reader(infile)\n",
    "    # skip header\n",
    "    next(reader, None)\n",
    "    # loop through rows\n",
    "    for rows in reader:\n",
    "        # retrieve information\n",
    "        filename = rows[0]\n",
    "        location = rows[1:5]\n",
    "        pneumonia = rows[5]\n",
    "        \n",
    "        # initialize our data\n",
    "        if filename not in used_filenames:\n",
    "            location_data = np.zeros((GRID_SIZE,GRID_SIZE,5))\n",
    "            \n",
    "            # default label to mean values\n",
    "            for i in range(4):\n",
    "                for j in range(4):\n",
    "                    y = ((i + 0.5) * CELL_WIDTH) - mu_center_y\n",
    "                    x = ((j + 0.5) * CELL_WIDTH) - mu_center_x\n",
    "                    location_data[i,j,:] = [0, x, y, 0, 0]\n",
    "            \n",
    "            used_filenames.append(filename)\n",
    "            locs = np.zeros((GRID_SIZE, GRID_SIZE))\n",
    "        else:\n",
    "            location_data = pneumonia_locations[filename]\n",
    "            \n",
    "        # if row contains pneumonia add label to dictionary\n",
    "        # the dictionary will contain the presence, x, y, h, w of the pneumonia for each pixel in a 16x16 grid\n",
    "        if pneumonia == '1':\n",
    "            # extract the location and cast them to floats\n",
    "            x,y,w,h = location\n",
    "            x = float(x)\n",
    "            y = float(y)\n",
    "            w = float(w)\n",
    "            h = float(h)\n",
    "            \n",
    "            # figure out the center of the ROI\n",
    "            center_x = x + (w / 2)\n",
    "            center_y = y + (h / 2)\n",
    "            \n",
    "            # figure out which pixel is in the center\n",
    "            x_pixel = int(center_x // (1024 // GRID_SIZE))\n",
    "            y_pixel = int(center_y // (1024 // GRID_SIZE))\n",
    "            locs[y_pixel, x_pixel] = 1\n",
    "            \n",
    "            # center the data\n",
    "            center_x = center_x - mu_center_x\n",
    "            center_y = center_y - mu_center_y\n",
    "            w = w - mu_w\n",
    "            h = h - mu_h\n",
    "            \n",
    "            # write the data to our list\n",
    "            location_data[y_pixel,x_pixel,0] = 1\n",
    "            location_data[y_pixel,x_pixel,1] = center_x\n",
    "            location_data[y_pixel,x_pixel,2] = center_y\n",
    "            location_data[y_pixel,x_pixel,3] = w\n",
    "            location_data[y_pixel,x_pixel,4] = h\n",
    "        else:\n",
    "            pass\n",
    "        \n",
    "        locations.append(location_data[:,:,1:])\n",
    "        pixel_means.append(np.mean(locs))\n",
    "        pneumonia_locations[filename] = location_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "foo = np.array(locations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-408.8788487"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.min(foo[:,:,0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28989, 4, 4, 4)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "foo.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save our dict to disk\n",
    "with open('yolo_labels_centered_4x4.p', 'wb') as handle:\n",
    "    pickle.dump(pneumonia_locations, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('yolo_labels_centered_4x4.p', 'rb') as handle:\n",
    "    b = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[   0.       , -373.8788487, -402.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       , -117.8788487, -402.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  138.1211513, -402.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  394.1211513, -402.2230589,    0.       ,\n",
       "            0.       ]],\n",
       "\n",
       "       [[   0.       , -373.8788487, -146.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       , -117.8788487, -146.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  138.1211513, -146.2230589,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  394.1211513, -146.2230589,    0.       ,\n",
       "            0.       ]],\n",
       "\n",
       "       [[   0.       , -373.8788487,  109.7769411,    0.       ,\n",
       "            0.       ],\n",
       "        [   1.       , -212.3788487,   19.2769411,    4.1546185,\n",
       "          194.8256359],\n",
       "        [   1.       ,  239.6211513,   97.2769411,  -19.8453815,\n",
       "          -25.1743641],\n",
       "        [   0.       ,  394.1211513,  109.7769411,    0.       ,\n",
       "            0.       ]],\n",
       "\n",
       "       [[   0.       , -373.8788487,  365.7769411,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       , -117.8788487,  365.7769411,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  138.1211513,  365.7769411,    0.       ,\n",
       "            0.       ],\n",
       "        [   0.       ,  394.1211513,  365.7769411,    0.       ,\n",
       "            0.       ]]])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pneumonia_locations[\"fffb2395-8edd-4954-8a89-ffe2fd329be3\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "37.441394898288664\n"
     ]
    }
   ],
   "source": [
    "print(1 / np.mean(pixel_means))"
   ]
  },
  {
   "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",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}