pototype for live tracking

demo.py

@@ -31,6 +31,8 @@ from utils.log import logger
 from utils.timer import Timer
 from utils.parse_config import parse_model_cfg
 import utils.datasets as datasets
+from pathlib import Path
+import tempfile
 from track import eval_seq
 
 
@@ -49,6 +51,8 @@ def track(opt):
     n_frame = 0
 
     logger.info('Starting tracking...')
+    
+    with tempfile.TemporaryDirectory() as tmpdirname:
         if os.path.isdir(opt.input_video):
             print('Use image sequence')
             dataloader = datasets.LoadImages(opt.input_video, opt.img_size)
@@ -58,16 +62,17 @@ def track(opt):
             frame_rate = dataloader.frame_rate 
         result_filename = os.path.join(result_root, 'results.txt')
 
-    frame_dir = None if opt.output_format=='text' else osp.join(result_root, 'frame')
+        frame_dir = None if opt.output_format=='text' else tmpdirname
         try:
             eval_seq(opt, dataloader, 'mot', result_filename,
-                 save_dir=frame_dir, show_image=False, frame_rate=frame_rate)
+                     save_dir=frame_dir, show_image=False, save_img=True, frame_rate=frame_rate)
         except Exception as e:
             logger.info(e)
 
         if opt.output_format == 'video':
-        output_video_path = osp.join(result_root, 'result.mp4')
+            name = 'result-' + Path(opt.input_video).stem + '.mp4'
-        cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg -c:v copy {}'.format(osp.join(result_root, 'frame'), output_video_path)
+            output_video_path = osp.join(result_root, name)
+            cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg -c:v mjpeg -q:v 1 {}'.format(frame_dir, output_video_path)
             os.system(cmd_str)
 
         

live_track.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 47,
+   "execution_count": 64,
    "id": "d1489f9f-328c-4812-9cdb-0a2dee44ae88",
    "metadata": {},
    "outputs": [],
@@ -28,7 +28,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 48,
+   "execution_count": 65,
    "id": "edec1b34-64ad-4610-856a-68d886a45142",
    "metadata": {},
    "outputs": [],
@@ -38,7 +38,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": 66,
    "id": "010bf567-8845-46d4-8500-883efce2d010",
    "metadata": {},
    "outputs": [],
@@ -51,7 +51,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": 67,
    "id": "8a413424-13c4-4bdc-825a-0aa6164e89e2",
    "metadata": {},
    "outputs": [],
@@ -67,7 +67,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 51,
+   "execution_count": 68,
    "id": "7b291b67-93ad-4b51-934a-dbaf095f7704",
    "metadata": {},
    "outputs": [],
@@ -103,7 +103,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 52,
+   "execution_count": 77,
+   "id": "c87f2b73-6109-4509-a0ce-62a690c44030",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'./OUT/embedding_test/track-test'"
+      ]
+     },
+     "execution_count": 77,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "result_frame_path"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 69,
    "id": "648faf4b-d692-473a-a99d-06b50a2e2261",
    "metadata": {},
    "outputs": [],
@@ -140,7 +161,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 53,
+   "execution_count": 70,
    "id": "a28ef404-2031-43cf-aeb1-357aa1be0934",
    "metadata": {},
    "outputs": [],
@@ -158,6 +179,28 @@
     "Load video file and get it's properties. Use that to calculate the dimension to fit the loaded model"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 74,
+   "id": "eb63b1c4-fdde-48e9-977e-21ceb10c5316",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(2, 512)"
+      ]
+     },
+     "execution_count": 74,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# array with shape n-components, n-features\n",
+    "reducer.components_.shape"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 59,
@@ -202,7 +245,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 75,
    "id": "c623aa17-5ce2-4948-9adf-d4c9a6d1ccd2",
    "metadata": {},
    "outputs": [
@@ -212,7 +255,7 @@
        "(-1.0, 1.0)"
       ]
      },
-     "execution_count": 61,
+     "execution_count": 75,
      "metadata": {},
      "output_type": "execute_result"
     },
@@ -242,7 +285,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": 76,
    "id": "e8952235-7e56-4606-858a-a9165b967726",
    "metadata": {},
    "outputs": [],
@@ -256,7 +299,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 63,
    "id": "f9cc6fd1-b9c2-4303-a21c-a193c6045526",
    "metadata": {},
    "outputs": [
@@ -273,6 +316,17 @@
      },
      "metadata": {},
      "output_type": "display_data"
+    },
+    {
+     "ename": "error",
+     "evalue": "OpenCV(4.7.0) /io/opencv/modules/imgproc/src/resize.cpp:4062: error: (-215:Assertion failed) !ssize.empty() in function 'resize'\n",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31merror\u001b[0m                                     Traceback (most recent call last)",
+      "Cell \u001b[0;32mIn[63], line 10\u001b[0m\n\u001b[1;32m      8\u001b[0m ret, frame \u001b[38;5;241m=\u001b[39m stream\u001b[38;5;241m.\u001b[39mread()\n\u001b[1;32m      9\u001b[0m \u001b[38;5;66;03m# scale down/up frame to fit tracker\u001b[39;00m\n\u001b[0;32m---> 10\u001b[0m frame \u001b[38;5;241m=\u001b[39m \u001b[43mcv2\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mresize\u001b[49m\u001b[43m(\u001b[49m\u001b[43mframe\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m(\u001b[49m\u001b[43mw\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mh\u001b[49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     13\u001b[0m \u001b[38;5;66;03m# letterbox as to have a constant size for the model\u001b[39;00m\n\u001b[1;32m     14\u001b[0m img, _, _, _ \u001b[38;5;241m=\u001b[39m datasets\u001b[38;5;241m.\u001b[39mletterbox(frame, height\u001b[38;5;241m=\u001b[39mopt\u001b[38;5;241m.\u001b[39mimg_size[\u001b[38;5;241m1\u001b[39m], width\u001b[38;5;241m=\u001b[39mopt\u001b[38;5;241m.\u001b[39mimg_size[\u001b[38;5;241m0\u001b[39m])\n",
+      "\u001b[0;31merror\u001b[0m: OpenCV(4.7.0) /io/opencv/modules/imgproc/src/resize.cpp:4062: error: (-215:Assertion failed) !ssize.empty() in function 'resize'\n"
+     ]
     }
    ],
    "source": [
@@ -366,10 +420,31 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 78,
    "id": "8f21e642-2320-4a7e-a0ea-af32e8d1a182",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "ffmpeg version 4.3 Copyright (c) 2000-2020 the FFmpeg developers\n",
+      "  built with gcc 7.3.0 (crosstool-NG 1.23.0.449-a04d0)\n",
+      "  configuration: --prefix=/opt/conda/conda-bld/ffmpeg_1597178665428/_h_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeh --cc=/opt/conda/conda-bld/ffmpeg_1597178665428/_build_env/bin/x86_64-conda_cos6-linux-gnu-cc --disable-doc --disable-openssl --enable-avresample --enable-gnutls --enable-hardcoded-tables --enable-libfreetype --enable-libopenh264 --enable-pic --enable-pthreads --enable-shared --disable-static --enable-version3 --enable-zlib --enable-libmp3lame\n",
+      "  libavutil      56. 51.100 / 56. 51.100\n",
+      "  libavcodec     58. 91.100 / 58. 91.100\n",
+      "  libavformat    58. 45.100 / 58. 45.100\n",
+      "  libavdevice    58. 10.100 / 58. 10.100\n",
+      "  libavfilter     7. 85.100 /  7. 85.100\n",
+      "  libavresample   4.  0.  0 /  4.  0.  0\n",
+      "  libswscale      5.  7.100 /  5.  7.100\n",
+      "  libswresample   3.  7.100 /  3.  7.100\n",
+      "\u001b[1;31mUnrecognized option 'crf'.\n",
+      "\u001b[0m\u001b[4;31mError splitting the argument list: \u001b[0m\u001b[4;31mOption not found\n",
+      "\u001b[0m"
+     ]
+    }
+   ],
    "source": [
     "! ffmpeg -i OUT/embedding_test/track-test/%04d.png -c:v libx264 -crf 10  OUT/embedding_test/track-test.mp4"
    ]

track.py

@@ -100,7 +100,8 @@ def eval_seq(opt, dataloader, data_type, result_filename, save_dir=None, save_im
         online_targets, frame_embeddings = tracker.update(blob, img0)
         online_tlwhs = []
         online_ids = []
-        for t in online_targets:
+        # for t in online_targets:
+        for t in tracker.tracked_stracks:
             tlwh = t.tlwh
             tid = t.track_id
             vertical = tlwh[2] / tlwh[3] > 1.6
@@ -118,7 +119,7 @@ def eval_seq(opt, dataloader, data_type, result_filename, save_dir=None, save_im
         if save_dir is not None:
             base_fn = os.path.join(save_dir, '{:05d}'.format(frame_id))
             if save_img:
-                cv2.imwrite(base_fn+'.jpg', online_im)
+                cv2.imwrite(base_fn+'.jpg', online_im, [cv2.IMWRITE_JPEG_QUALITY, 100])
             if save_figures:
                 for i, fe in enumerate(frame_embeddings):
                     tlwh, curr_feat = fe