README_Tony.md

SparseTrack

0. Build docker image and continue from step 6

docker build -t <image_name> .

OR

docker pull 12labs/sparsetrack

1. Clone repo

git clone https://github.com/swmoon00/SparseTrack.git

2. Install apt packages (with sudo)

apt install libboost-all-dev
apt install libpython3-dev

3. Configure conda
   • The python version should be matched with /usr/bin/python3 --version e.g.

conda create -n sparsetrack python=3.8.10

4. Install pbcvt package and copy library file Edit CMakeLists.txt file as in SparseTrack instructions, and compile with make

5. Install cython_bbox

   • There is a bug in cython_bbox package.
   • Install patched version

pip install git+https://github.com/swmoon00/cython_bbox.git

6. Download model weights

   • https://drive.google.com/file/d/1P4mY0Yyd3PPTybgZkjMYhFri88nTmJX5/view
   • Change train.init_checkpoint option in mot17_track_cfg.py file

7. Run demo

python demo_track.py --config-file mot17_track_cfg.py --video-input videos/palace.mp4