An open API service indexing awesome lists of open source software.

https://github.com/adept-thu/CAMO-MOT

Official code of CAMO-MOT
https://github.com/adept-thu/CAMO-MOT

camera-lidar multiple-object-tracking

Last synced: about 1 year ago
JSON representation

Official code of CAMO-MOT

Awesome Lists containing this project

README

          

# CAMO-MOT

This is the official repo release of the paper [CAMO-MOT: Combined Appearance-Motion Optimization for 3D Multi-Object Tracking with Camera-LiDAR Fusion](https://arxiv.org/abs/2209.02540).

## News

- 2022-09-08. CAMO-MOT is released on [arXiv](https://arxiv.org/abs/2209.02540):slightly_smiling_face:.
- 2022-08-04. We rank **first** among all methods on **nuScenes** Dataset for [Tracking](https://www.nuscenes.org/tracking?externalData=all&mapData=all&modalities=Any):blush:.
- 2022-08-03. We rank **4th** among all methods on **KITTI** Dataset for [MOT](http://www.cvlibs.net/datasets/kitti/eval_tracking.php):grinning:.

## Results

### Multi-object tracking(on nuScenes test set)

Method | AMOTA | AMOTP
--------------- |:--------:|:--------:
CAMO-MOT | 0.753 | 0.472


You can find detailed results on nuScenes **test** set on this [website](https://eval.ai/web/challenges/challenge-page/476/leaderboard/1321).
Or you can view the accuracy trend of MOT algorithms on this [website](https://paperswithcode.com/sota/3d-multi-object-tracking-on-nuscenes)

### Multi-object tracking(on nuScenes val set)

Tracker | AMOTA | AMOTP
--------------- |:--------:|:--------:
CAMO-MOT | 0.763 | 0.527

On nuScenes, we use [BEVFusion](https://github.com/mit-han-lab/bevfusion) and [FocalConv](https://github.com/dvlab-research/FocalsConv) as our detectors.

### Multi-object tracking(on KITTI test)

Category | HOTA (%) | MOTA (%) | MOTP (%)| MT (%) | ML (%) | IDS | FRAG | FP | FN
--------------- |:--------:|:--------:|:-------:|:------:|:------:|:---:|:----:|:----:|:-----:
*Car* | 79.99 | 90.38 | 85.00 | 84.46 | 7.54 | 30 | 156 | 2337 | 942
*Pedestrian* | 44.77 | 52.48 | 64.50 | 35.40 | 25.77 | 152 | 1133 | 8325 | 2525

You can find detailed results on KITTI **test** set on this [website](http://www.cvlibs.net/datasets/kitti/eval_tracking_detail.php?result=b3be646ab7ac4935ad15cb81cc1e12a6d8db4983).

### Multi-object tracking(on KITTI val)

Category | HOTA (%) | MOTA (%) | IDS | FP | FN
--------------- |:--------:|:--------:|:---:|:----:|:-----:
*Car* | 82.91 | 91.96 | 1 | 302 | 371
*Pedestrian* | 50.99 | 64.75 | 70 | 2240 | 1140

On KITTI, we use [PointGNN](https://github.com/WeijingShi/Point-GNN) as our detector.

## License

`CAMO-MOT` is released under the `MIT` license.

## Acknowledgement

In the detection part, many thanks to the following open-source projects:
- [CenterPoint](https://github.com/tianweiy/CenterPoint)
- [FocalConv](https://github.com/dvlab-research/FocalsConv)
- [BEVFusion](https://github.com/mit-han-lab/bevfusion)
- We especially thank Yukang@yukang2017([FocalConv](https://github.com/dvlab-research/FocalsConv)) for his help.

In the tracking part, many thanks to the following open-source projects:
- [DEFT](https://github.com/MedChaabane/DEFT)
- [EagerMOT](https://github.com/aleksandrkim61/EagerMOT)
- [PC3T](https://github.com/TRI-ML/permatrack)

## Citation
If you find our paper useful for you, please consider cite us by:grin::
```
@ARTICLE{10164676,
author={Wang, Li and Zhang, Xinyu and Qin, Wenyuan and Li, Xiaoyu and Gao, Jinghan and Yang, Lei and Li, Zhiwei and Li, Jun and Zhu, Lei and Wang, Hong and Liu, Huaping},
journal={IEEE Transactions on Intelligent Transportation Systems},
title={CAMO-MOT: Combined Appearance-Motion Optimization for 3D Multi-Object Tracking With Camera-LiDAR Fusion},
year={2023},
volume={},
number={},
pages={1-16},
doi={10.1109/TITS.2023.3285651}}
```