{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Duke MTMC BBox Visualize\n",
    "\n",
    "- read in all images, resized\n",
    "- mean images\n",
    "- camera, ID, frame, left, top, width, height, wx, wy, feetX, feetyY\n",
    "- read in all bboxes\n",
    "- draw normalized bboxes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import os\n",
    "from os.path import join\n",
    "import math\n",
    "from glob import glob\n",
    "from random import randint\n",
    "\n",
    "import cv2 as cv\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from PIL import Image, ImageDraw\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy.io as sio\n",
    "import h5py\n",
    "from pathlib import Path\n",
    "from sklearn import preprocessing\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "\n",
    "import sys\n",
    "sys.path.append('/work/megapixels_dev/megapixels/')\n",
    "from app.settings import app_cfg as cfg\n",
    "from app.utils import file_utils, im_utils\n",
    "from app.models.bbox import BBox"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "fp_det_trainval = '/data_store/datasets/people/duke_mtmc/dataset/trainvalRaw.mat'\n",
    "fp_dir_processed = '/data_store/datasets/people/duke_mtmc/processed/'\n",
    "fp_dir_vids = '/data_store/datasets/people/duke_mtmc/dataset/videos/'\n",
    "fp_out_bboxes = '/data_store/datasets/people/duke_mtmc/processed/duke_mtmc_bboxes_01.png'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<KeysViewHDF5 ['trainData']>\n"
     ]
    }
   ],
   "source": [
    "cam_dets = {}\n",
    "with h5py.File(fp_det_trainval, 'r') as f:\n",
    "  print(f.keys())\n",
    "  for k, v in f.items():\n",
    "    cam_dets[k] = np.array(v)\n",
    "train_data = cam_dets['trainData']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "mat_keys = ['camera', 'id', 'frame', 'left', 'top', 'width', 'height', 'wx', 'wy', 'feetX', 'feetY']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "detections = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7eaff9a503e8426380c71e84a262f448",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=4994653), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "total = len(train_data[0])\n",
    "for i in tqdm(range(0, total)):\n",
    "  camera_str = str(int(train_data[0][i]))\n",
    "  obj = {\n",
    "    'camera': int(train_data[0][i]),\n",
    "    'id': int(train_data[1][i]),\n",
    "    'frame': int(train_data[2][i]),\n",
    "    'left': int(train_data[3][i]),\n",
    "    'top': int(train_data[4][i]),\n",
    "    'width': int(train_data[5][i]),\n",
    "    'height': int(train_data[6][i]),\n",
    "    'wx': train_data[7][i],\n",
    "    'wy': train_data[8][i],\n",
    "    'feetX': train_data[9][i],\n",
    "    'feetY': train_data[10][i]\n",
    "  }\n",
    "  if camera_str not in detections.keys():\n",
    "    detections[camera_str] = []\n",
    "  detections[camera_str].append(obj)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Average of Video"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5ee08cb23c844e009d5ca5bd2ef39d09",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0f4bafaeaed7461d8ebba036890db289",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c1a936c427b0425eb910be1d7115669b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9a0e24e13f0c44f4a2574ad5ae3be63c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d3bdb03e99804938b655e9083d5823df",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "43ce9c80dddd42239e718dbd40ba1e07",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=9), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c445c047d97e4825a41b163782f795b5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=9), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "77584ca01ac5496689dc6d2955c3d3ed",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "for cam in range(1,9):\n",
    "  frames = []\n",
    "  # skip frames\n",
    "  dec = 480\n",
    "  fps_video_cam = glob(join(fp_dir_vids, f'camera{cam}/*.mp4'))\n",
    "\n",
    "  for fp_video_cam in tqdm(fps_video_cam):\n",
    "    pos = 0\n",
    "    vid = cv.VideoCapture(fp_video_cam)\n",
    "    nframes = int(vid.get(cv.CAP_PROP_FRAME_COUNT))\n",
    "\n",
    "    # read video frames\n",
    "    while vid.isOpened():\n",
    "      pos += dec\n",
    "      if pos >= nframes:\n",
    "        break\n",
    "      vid.set(cv.CAP_PROP_POS_FRAMES, pos)\n",
    "      read, frame = vid.read()\n",
    "      if not read:\n",
    "        break\n",
    "      frames.append(frame)\n",
    "  \n",
    "  vid.release()\n",
    "  # save average for each camera\n",
    "  im_mean = np.mean(frames, axis=0).astype(np.uint8)\n",
    "  fp_out_cam_avg = join(fp_dir_processed, f'duke_mtmc_cam{cam}_average.png')\n",
    "  cv.imwrite(fp_out_cam_avg, im_mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.imshow(cv.cvtColor(im_mean, cv.COLOR_BGR2RGB))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "37f7ad3c034f4b49ade20a711f76ce26",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=1300168), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# create list of bboxes\n",
    "dim = (1920, 1080)\n",
    "bboxes = []\n",
    "for d in tqdm(detections[str(cam)]):\n",
    "  bboxes.append(BBox.from_xywh_dim(d['left'], d['top'], d['width'], d['height'], dim))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fcdf5a38255540c38c544ccf6b459fa1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=1300168), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "im_saliency = np.zeros([dim[1], dim[0]]).astype(np.float32)\n",
    "for i, bbox in enumerate(tqdm(bboxes)):\n",
    "  bbox = bbox.to_dim(dim)\n",
    "  pt1, pt2 = tuple(map(int, bbox.pt_tl)), tuple(map(int, bbox.pt_br))\n",
    "  im_saliency[pt1[1]:pt2[1], pt1[0]:pt2[0]] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im_saliency_mean = (im_saliency /  (np.max(im_saliency) / 1)).astype(np.float32)\n",
    "im_saliency_mean = 255 * im_saliency_mean\n",
    "im_saliency_mean = np.clip(im_saliency_mean, 0, 255).astype(np.uint8)\n",
    "im_cm = cv.applyColorMap(im_saliency_mean, cv.COLORMAP_JET)\n",
    "im_cm = cv.blur(im_cm, ksize=(13, 13))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.imshow(cv.cvtColor((im_cm).astype(np.uint8), cv.COLOR_BGR2RGB))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save files\n",
    "fp_out_saliency = join(fp_dir_processed, f'duke_mtmc_saliency_cam{cam}.png')\n",
    "cv.imwrite(fp_out_saliency, im_saliency_mean)\n",
    "#im_saliency_mean = cv.blur(im_saliency_mean, ksize=(5,5))\n",
    "#fp_out_saliency_blur = join(fp_dir_processed, f'duke_mtmc_saliency_cam{cam}_blur.png')\n",
    "cv.imwrite(fp_out_saliency_blur, im_saliency_mean)\n",
    "# wave"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'im_cm' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-8-2462e63f80c0>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# preview overlay (not for production)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0malpha\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0.5\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mim_comp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maddWeighted\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mim_mean\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0malpha\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mim_cm\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0malpha\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimshow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcvtColor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mim_comp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCOLOR_BGR2RGB\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'im_cm' is not defined"
     ]
    }
   ],
   "source": [
    "# preview overlay (not for production)\n",
    "alpha = 0.5\n",
    "im_comp = cv.addWeighted(im_mean, alpha, im_cm, 1 - alpha, 0)\n",
    "plt.imshow(cv.cvtColor(im_comp, cv.COLOR_BGR2RGB))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import blend_modes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "blend_modes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Old Viz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i, bbox in enumerate(tqdm(bboxes)):\n",
    "  if i % 60:\n",
    "    continue\n",
    "  bbox = bbox.to_dim(size_draw)\n",
    "  pt1, pt2 = tuple(map(int, bbox.pt_tl)), tuple(map(int, bbox.pt_br))\n",
    "  im_draw_next = im_bg.copy()\n",
    "  cv.rectangle(im_draw_next, pt1, pt2, clr, stroke_weight, cv.LINE_AA)\n",
    "  im_bg = cv.addWeighted(im_draw_next, alpha, im_bg, 1 - alpha, 0)\n",
    "  if i > limit:\n",
    "    break\n",
    "\n",
    "cv.imwrite(fp_out_bboxes, im_bg)\n",
    "plt.imshow(cv.cvtColor(im_bg, cv.COLOR_BGR2RGB))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# draw it\n",
    "clr = (255, 255, 255)\n",
    "size_orig = (5184, 3456)\n",
    "scale = size_orig[0]/1920\n",
    "size_draw = (int(size_orig[0]/scale), int(size_orig[1]/scale))\n",
    "im_draw = np.zeros([size_draw[1], size_draw[0], 3]).astype(np.uint8)\n",
    "im_bg = im_draw.copy()\n",
    "limit = 2000000\n",
    "stroke_weight = 1\n",
    "alpha = 0.125\n",
    "\n",
    "for i, roi in enumerate(tqdm(rois.items())):\n",
    "  fname, bbox = roi  # expand\n",
    "  bbox = bbox.to_dim(size_draw)\n",
    "  pt1, pt2 = tuple(map(int, bbox.pt_tl)), tuple(map(int, bbox.pt_br))\n",
    "  im_draw_next = im_bg.copy()\n",
    "  cv.rectangle(im_draw_next, pt1, pt2, clr, stroke_weight, cv.LINE_AA)\n",
    "  im_bg = cv.addWeighted(im_draw_next, alpha, im_bg, 1 - alpha, 0)\n",
    "  if i > limit:\n",
    "    break\n",
    "\n",
    "\n",
    "plt.imshow(cv.cvtColor(im_bg, cv.COLOR_BGR2RGB))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Average Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im_comp = np.empty_like(im_utils.resize(cv.imread(fp_im), width=1920) / 255)\n",
    "for fp_im in tqdm(fp_ims[:100]):\n",
    "  im = im_utils.resize(cv.imread(fp_im), width=1920) / 255\n",
    "  im_comp += im"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im_mean = np.mean(ims, axis=0).astype(np.uint8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.imshow(im_mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "a = np.array([.1, .2])\n",
    "b = np.array([.1, .2])\n",
    "c = a + b\n",
    "print(c)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im = im.astype(np.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "megapixels",
   "language": "python",
   "name": "megapixels"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}