json _mask_reading.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json, os, glob\n",
    "from PIL import Image, ImageDraw\n",
    "import numpy as np\n",
    "from PIL.ImageColor import getrgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_dict = {'NFL':13, 'GCL+IPL':5, 'INL':17, 'OPL':2,'ONL':22,'Elipsoid Zone':9,'RPE':3,'Choroid':31}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "json_file = '/home/khamituly/Documents/Diploma work/labels_8_layer_2022-06-08-03-14-52.json'\n",
    "original_images = '/home/khamituly/Documents/Diploma work/dataset_manual/temp_health/X/'\n",
    "result_save = '/home/khamituly/Documents/Diploma work/dataset_manual/temp_health/Y_test/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_labels = json.load(open(json_file))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['06 OS - Copy.jpeg', '108 OD - Copy.jpeg', '22 OD - Copy.jpeg', '26 OD - Copy.jpeg', '31 OD - Copy.jpeg', '33 OD - Copy.jpeg', '34 OS - Copy.jpeg', '50 OD - Copy.jpeg', '56 OS - Copy.jpeg', '64 OD - Copy.jpeg'])\n"
     ]
    }
   ],
   "source": [
    "rawlist = all_labels.keys()\n",
    "print(all_labels.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def get_data(data, X,Y,L):\n",
    "def get_data(data):\n",
    "\n",
    "    X = []; Y = []; L=[] #pre-allocate lists to fill in a for loop\n",
    "    for k in data['regions']: #cycle through each polygon\n",
    "        # get the x and y points from the dictionary\n",
    "        X.append(data['regions'][k]['shape_attributes']['all_points_x'])\n",
    "        Y.append(data['regions'][k]['shape_attributes']['all_points_y'])\n",
    "        L.append(data['regions'][k]['region_attributes']['label'])\n",
    "    return Y,X,L #image coordinates are flipped relative to json coordinates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_mask(X, Y, nx, ny, L, class_dict):\n",
    "    # get the dimensions of the image\n",
    "    mask = np.zeros((nx,ny))\n",
    "\n",
    "    for y,x,l in zip(X,Y,L):\n",
    "        # the ImageDraw.Draw().polygon function we will use to create the mask\n",
    "        # requires the x's and y's are interweaved, which is what the following\n",
    "        # one-liner does\n",
    "        polygon = np.vstack((x,y)).reshape((-1,),order='F').tolist()\n",
    "\n",
    "        # create a mask image of the right size and infill according to the polygon\n",
    "        size = (nx, ny)\n",
    "        if nx>ny:\n",
    "           x,y = y,x\n",
    "           img = Image.new(\"L\",  (nx, ny), 0)\n",
    "        elif ny>nx:\n",
    "           img = Image.new(\"L\", (ny,nx), 0)\n",
    "        else:\n",
    "           img = Image.new(\"L\", (nx, ny),  0)\n",
    "\n",
    "        ImageDraw.Draw(img).polygon(polygon, fill = 1, outline = 1)\n",
    "        # turn into a numpy array\n",
    "        m = np.flipud(np.rot90(np.array(img)))\n",
    "\n",
    "        try:\n",
    "            mask[m==1] = class_dict[l]\n",
    "        except:\n",
    "            mask[m.T==1] = class_dict[l]  \n",
    "\n",
    "         # mask[m==1] = class_dict[l]\n",
    "\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rescale(dat,mn,mx):\n",
    "    '''\n",
    "    rescales an input dat between mn and mx\n",
    "    '''\n",
    "    m = min(dat.flatten())\n",
    "    M = max(dat.flatten())\n",
    "    return (mx-mn)*(dat-m)/(M-m)+mn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [],
   "source": [
    "for name in rawlist:\n",
    "    \n",
    "    new_path = original_images+name\n",
    "    image = Image.open(new_path)\n",
    "    nx, ny, nz = np.shape(image)\n",
    "    \n",
    "    X, Y, L = get_data(all_labels[name])\n",
    "\n",
    "    mask = get_mask(X, Y, nx, ny, L, class_dict)\n",
    "\n",
    "    mask = Image.fromarray(rescale(mask,0,255)).convert('RGB')\n",
    "    mask.save(result_save+name.replace('imagery','labels'), format='PNG')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = cv2.imread(result_save+'33 OD - Copy.jpeg')\n",
    "img = cv2.resize(img,(640,640), interpolation = cv2.INTER_NEAREST)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 3, 4, 5, 6, 7, 8])"
      ]
     },
     "execution_count": 286,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder\n",
    "labelencoder = LabelEncoder()\n",
    "n, h, w = img.shape\n",
    "train_masks_reshaped = img.reshape(-1,1)\n",
    "train_masks_reshaped_encoded = labelencoder.fit_transform(train_masks_reshaped)\n",
    "train_masks_encoded_original_shape = train_masks_reshaped_encoded.reshape(n, h, w)\n",
    "\n",
    "np.unique(train_masks_encoded_original_shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 3, 4, 5, 6, 7, 8])"
      ]
     },
     "execution_count": 287,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "train_masks_input = np.expand_dims(train_masks_encoded_original_shape, axis=3)\n",
    "np.unique(train_masks_input)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  0,  16,  24,  41,  74, 106, 139, 180, 255], dtype=uint8)"
      ]
     },
     "execution_count": 288,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Capture mask/label info as a list\n",
    "train_masks = [] \n",
    "for directory_path in glob.glob(result_save):\n",
    "    for mask_path in glob.glob(os.path.join(directory_path, \"*.jpeg\")):\n",
    "        mask = cv2.imread(mask_path, 0)       \n",
    "        mask = cv2.resize(mask, (640,640), interpolation = cv2.INTER_NEAREST)  #Otherwise ground truth changes due to interpolation\n",
    "        train_masks.append(mask)\n",
    "        \n",
    "#Convert list to array for machine learning processing          \n",
    "train_masks = np.array(train_masks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[  0  16  24  41  74 106 139 180 255]   label0\n",
      "[  0  16  24  41  74 106 139 180 255]   label1\n",
      "[  0  16  24  41  74 106 139 180 255]   label2\n",
      "[  0  16  24  41  74 106 139 180 255]   label3\n",
      "[  0  16  24  41  74 106 139 180 255]   label4\n",
      "[  0  16  24  41  74 106 139 180 255]   label5\n",
      "[  0  16  24  41  74 106 139 180 255]   label6\n",
      "[  0  16  24  41  74 106 139 180 255]   label7\n",
      "[  0  16  24  41  74 106 139 180 255]   label8\n",
      "[  0  16  24  41  74 106 139 180 255]   label9\n"
     ]
    }
   ],
   "source": [
    "for i in range(10):\n",
    "    print(np.unique(train_masks[i]),'  label'+str(i))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f53e499fc40>"
      ]
     },
     "execution_count": 291,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "plt.imshow(train_masks[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f53e494d880>"
      ]
     },
     "execution_count": 292,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "plt.imshow(train_masks[3])"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
  },
  "kernelspec": {
   "display_name": "Python 3.8.10 64-bit",
   "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": "3.8.10"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}