Derived from
https://github.com/nwojke/deep_sort.git
This repository is an update to tensorflow v2 for the famous deep_sort project. It is necessary to convert the model given in the original repository to a v2 format for tensorflow. This is accomplished using the convert.py script included. You will need to get the original format model from the nwojke/deep_sort repository to test the script. The final v2 saved_model format is included with this repository.
The test program is hard coded for a MOT 16 Benchmark sequence. We assume resources have been extracted to the repository root directory and the MOT16 benchmark data is in ./MOT16:
The v2model.py script will run the deep_sort test on the MOT16 files. The script has been streamlined to reduce the code base, but the underlying functionality should be identical to the original deep_sort program.
Model development and training code has been omitted from this repository, it is only concerned with run time functionality.
package versions used:
- opencv-python 4.4.0.42
- tensorflow 2.6.0
- numpy 1.19.5
- sklearn 0.0
- scikit-learn 0.22.2
- scipy 1.4.1
Revised: Added utils/linear_assignment.py to replace deprecated function in sklearn
Added interface.py for C++ link
If you find this repo useful in your research, please consider citing the following papers:
@inproceedings{Wojke2017simple,
title={Simple Online and Realtime Tracking with a Deep Association Metric},
author={Wojke, Nicolai and Bewley, Alex and Paulus, Dietrich},
booktitle={2017 IEEE International Conference on Image Processing (ICIP)},
year={2017},
pages={3645--3649},
organization={IEEE},
doi={10.1109/ICIP.2017.8296962}
}
@inproceedings{Wojke2018deep,
title={Deep Cosine Metric Learning for Person Re-identification},
author={Wojke, Nicolai and Bewley, Alex},
booktitle={2018 IEEE Winter Conference on Applications of Computer Vision (WACV)},
year={2018},
pages={748--756},
organization={IEEE},
doi={10.1109/WACV.2018.00087}
}