File size: 13,300 Bytes
2b7d279 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 | ## Beyond the Survey: A Systematic Empirical Study of Detection and Association in Visual MOT
Official implementation repository for the paper **"Beyond the Survey: A Systematic Empirical Study of Detection and Association in Visual MOT"**, accepted for publication in *Artificial Intelligence Review*.
> **Note:** This repository also includes the official Python implementation for:
> **"Adaptive Confidence Threshold for ByteTrack in Multi-Object Tracking"** (*ICCAIS 2023*, [arXiv 2312.01650](https://arxiv.org/abs/2312.01650)) at this folder [bytetrack](trackers/bytetrack).
---
### Overview
This repository provides tracking implementations for algorithms evaluated in our study: **SORT, DeepSORT, MOTDT, FairMOT, ByteTrack (AdaptByteTrack), OCSORT, DeepOCSORT, and VisualRFS**. We provide CMC (Camera Motion Compensation) for ByteTrack (AdaptByteTrack), VisualRFS, and OC-SORT. Evaluation Scores/Metrics are:
- [CLEAR](https://link.springer.com/article/10.1155/2008/246309), [IDF1](https://arxiv.org/pdf/1609.01775), [HOTA](https://arxiv.org/abs/2009.07736) evaluation available at [JonathonLuiten/TrackEval](https://github.com/JonathonLuiten/TrackEval).
- Tracking Effort Measure [TEM](https://arxiv.org/abs/2212.08536) ([vpulab/MOT-evaluation](https://github.com/vpulab/MOT-evaluation)).
For other evaluated methods, please clone their respective repositories and follow the authors' original execution instructions.
---
### Datasets & Detection Outputs
#### Pre-extracted Detections can be downloaded from Hugging Face ([linhmv/VisualMOT](https://huggingface.co/datasets/linhmv/VisualMOT)).
We provide extracted detection files with confidence scores $[0, 1]$ in the ([`./dets/`](./dets/)) directory:
| Detector | Venue / Source | Paper / Link |
|:-------------------------------| :--- | :--- |
| POI: `detector_poi` | ECCV 2016 | [arXiv:1610.06136](https://arxiv.org/pdf/1610.06136) |
| JDE: `detector_jde` | ECCV 2020 | [arXiv:1909.12605](https://arxiv.org/abs/1909.12605) |
| TraDeS: `detector_trades` | CVPR 2021 | [arXiv:2103.08808](https://arxiv.org/abs/2103.08808) |
| FairMOT: `detector_fairmot128` | IJCV 2021 | [arXiv:2004.01888](https://arxiv.org/abs/2004.01888) |
| GSDT: `detector_gsdt` | ICRA 2021 | [arXiv:2006.13164](https://arxiv.org/abs/2006.13164) |
| CSTrack: `detector_cstrack` | TIP 2022 | [arXiv:2010.12138](https://arxiv.org/abs/2010.12138) |
| YOLOX: `detector_bytetrack` | ECCV 2022 | [arXiv:2110.06864](https://arxiv.org/abs/2110.06864) |
| YOLOv11: `detectors_yolov11` | arXiv 2024 | [arXiv:2410.17725](https://arxiv.org/abs/2410.17725) |
- POI: Relies on [ETHZ](https://ieeexplore.ieee.org/document/4587581), [Caltech Pedestrian](https://ieeexplore.ieee.org/abstract/document/5206631), and a self-collected surveillance dataset.
- TraDeS: Utilizes a [CrowdHuman](https://arxiv.org/abs/1805.00123) pre-trained model for 2D tracking alongside the [MOTChallenge](https://motchallenge.net/) dataset.
- JDE, FairMOT, CSTrack, GSDT: Fine-tuned on the "[Mix of Six](https://github.com/Zhongdao/Towards-Realtime-MOT/blob/master/DATASET_ZOO.md)" dataset, which combines [Caltech Pedestrian](https://ieeexplore.ieee.org/abstract/document/5206631), [CityPersons](https://arxiv.org/abs/1702.05693), [ETHZ](https://ieeexplore.ieee.org/document/4587581), [MOTChallenge](https://motchallenge.net/), [CUHK-SYSU](https://arxiv.org/abs/1604.01850), and [PRW](https://arxiv.org/abs/1604.02531).
- YOLOX: Trained on a combination of [MOTChallenge](https://motchallenge.net/), [CrowdHuman](https://arxiv.org/abs/1805.00123), [CityPersons](https://arxiv.org/abs/1702.05693), and [ETHZ](https://ieeexplore.ieee.org/document/4587581).
- YOLOv11: `YOLO11x`, Only trained on [COCO Detection](https://cocodataset.org/#overview) Dataset
#### Evaluation Datasets
Dataset Name | Year | Source|
| :--- |:---------------------------------------------------------------------------------------------| :--- |
| [MOTChallenge](https://motchallenge.net/)| [2016](https://arxiv.org/abs/1603.00831), [2017](https://motchallenge.net/data/MOT17/), [2020](https://arxiv.org/abs/2003.09003) |
|[DanceTrack](https://arxiv.org/abs/2111.14690)| CVPR 2022 |[DanceTrack/DanceTrack](https://github.com/DanceTrack/DanceTrack)
|[SportsMOT](https://arxiv.org/abs/2304.05170)| ICCV 2023 |[MCG-NJU/SportsMOT](https://github.com/MCG-NJU/SportsMOT)
|[CrowdTrack](https://arxiv.org/abs/2507.02479)| arXiv 2025 | [loseevaya/CrowdTrack](https://github.com/loseevaya/CrowdTrack)
---
### Evaluated Tracking Algorithms
#### 1. Analytical Data Association
*Hand-crafted motion & appearance models*
| Method | Ref. Index | Year | Source Code |
|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| :---: |:------------:| :--- |
| **[SORT](https://arxiv.org/abs/1602.00763)** | [23] | ICIP 2016 | [abewley/sort](https://github.com/abewley/sort) |
| **[DeepSORT](https://arxiv.org/abs/1703.07402)** | [27] | ICIP 2017 | [nwojke/deep_sort](https://github.com/nwojke/deep_sort) |
| **[MOTDT](https://arxiv.org/abs/1809.04427)** | [28] | ICME 2018 | [longcw/MOTDT](https://github.com/longcw/MOTDT) |
| **[FairMOT](https://arxiv.org/abs/2004.01888)** | [29] | IJCV 2021 | [ifzhang/FairMOT](https://github.com/ifzhang/FairMOT) |
| **[ByteTrack](https://arxiv.org/abs/2110.06864)** | [24] | ECCV 2022 | [FoundationVision/ByteTrack](https://github.com/FoundationVision/ByteTrack) |
| **[AdaptByteTrack](https://arxiv.org/abs/2312.01650)** | [75] | ICCAIS 2023 | [linh-gist/AdaptConfByteTrack](https://github.com/linh-gist/AdaptConfByteTrack) |
| **[OCSORT](https://arxiv.org/abs/2203.14360)** | [25] | CVPR 2023 | [noahcao/OC_SORT](https://github.com/noahcao/OC_SORT) |
| **[DeepOCSORT](https://arxiv.org/abs/2302.11813)** | [30] | ICIP 2023 | [gerardmaggiolino/deep-oc-sort](https://github.com/gerardmaggiolino/deep-oc-sort) |
| **[StrongSORT](https://arxiv.org/abs/2202.13514)** | [32] | TMM 2023 | [dyhBUPT/StrongSORT](https://github.com/dyhBUPT/StrongSORT) |
| **[VisualRFS](https://arxiv.org/abs/2407.08872)** | [1] | PR 2024 | [linh-gist/VisualRFS](https://github.com/linh-gist/VisualRFS) |
| **[HybridSORT](https://arxiv.org/abs/2308.00783)** | [31] | AAAI 2024 | [ymzis69/HybridSORT](https://github.com/ymzis69/HybridSORT) |
| **[TrackTrack](https://openaccess.thecvf.com/content/CVPR2025/html/Shim_Focusing_on_Tracks_for_Online_Multi-Object_Tracking_CVPR_2025_paper.html)** | [26] | CVPR 2025 | [kamkyu94/TrackTrack](https://github.com/kamkyu94/TrackTrack) |
#### 2. Deep Learning Data Association
*Learned feature-based association*
| Method | Ref. Index | Year | Source Code |
|:-----------------------------------------------------------------------------------------------------| :---: |:-----------:| :--- |
| **[SUSHI](https://arxiv.org/abs/2212.03038)** | [33] | CVPR 2023 | [dvl-tum/sushi](https://github.com/dvl-tum/sushi) |
| **[LTTrack](https://ieeexplore.ieee.org/abstract/document/10536914)** | [34] | TCSVT 2024 | [linjiaping1/LTTrack](https://github.com/linjiaping1/LTTrack) |
| **[LG-MOT](https://arxiv.org/abs/2406.04844)** | [35] | TCSVT 2025 | [weslee88524/lg-mot](https://github.com/weslee88524/lg-mot) |
#### 3. End-to-End (E2E) Data Association
*Joint detection and association learning*
| Method | Ref. Index | Year | Source Code |
|:------------------------------------------------------------------------------| :---: |:----------:| :--- |
| **[MOTR](https://arxiv.org/pdf/2105.03247)** | [17] | ECCV 2022 | [megvii-research/MOTR](https://github.com/megvii-research/MOTR) |
| **[MeMOTR](https://arxiv.org/abs/2307.15700)** | [41] | ICCV 2023 | [mcg-nju/memotr](https://github.com/mcg-nju/memotr) |
| **[MOTIP](https://arxiv.org/abs/2403.16848)** | [42] | CVPR 2025 | [MCG-NJU/MOTIP](https://github.com/MCG-NJU/MOTIP) |
| **[CO-MOT](https://arxiv.org/abs/2305.12724)** | [43] | ICLR 2025 | [BingfengYan/CO-MOT](https://github.com/BingfengYan/CO-MOT) |
| **[SambaMOTR](https://arxiv.org/abs/2410.01806)** | [39] | ICLR 2025 | [mattiasegu/sambamotr](https://github.com/mattiasegu/sambamotr) |
### Usage
1. **Set Up Python Environment**
- Create a `conda` Python environment and activate it:
```sh
conda create --name virtualenv python==3.8.0
conda activate virtualenv
```
- lone this repository recursively to have pybind11
```sh
git clone --recursive https://github.com/linh-gist/AdaptConfByteTrack.git
```
- Install Packages
```sh
numpy==1.23.1
opencv-python==4.9.0.80
loguru==0.7.2
scipy==1.10.1
lap==0.5.12
cython_bbox==0.1.5
matplotlib==3.5.3
filterpy==1.4.5
motmetrics==1.4.0
openpyxl==3.1.5
pycocotools==2.0.7
tabulate==0.9.0
# git clone https://github.com/JonathonLuiten/TrackEval.git
# cd TrackEval, python setup.py build develop
```
2. **Prepare Data**
- Datasets:
- MOT16, MOT17, MOT20, DanceTrack, SportsMOT, CrowdTrack
- You can also run with your custom dataset but need a detector
3. **Run the Tracking Demo**
- Change parameters in `make_parser()` in `track.py` such as `use_gmc`, `data_dir` (MOTChallenge GT data)
- Run `python track.py`
### Citation
If you find this project useful in your research, please consider citing by:
```
@article{van2026beyond,
title={Beyond the Survey: A Systematic Empirical Study of Detection and Association in Visual MOT},
author={Linh Van Ma and Juhua Hu and Wei Cheng and Unse Fatima and Moongu Jeon},
booktitle={Artificial Intelligence Review},
year={2026},
publisher={Springer}
}
@inproceedings{van2023adaptive,
title={Adaptive Confidence Threshold for ByteTrack in Multi-Object Tracking},
author={Linh Van Ma and Muhammad Ishfaq Hussain and JongHyun Park and Jeongbae Kim and Moongu Jeon},
booktitle={2023 12th International Conference on Control, Automation and Information Sciences (ICCAIS)},
pages={370--374},
year={2023},
organization={IEEE}
}
```
### Acknowledgement
A part of the code is borrowed from [SORT](https://github.com/abewley/sort), [DeepSORT](https://github.com/nwojke/deep_sort), [MOTDT](https://arxiv.org/abs/1809.04427), [FairMOT](https://arxiv.org/abs/2004.01888), [ByteTrack](https://arxiv.org/abs/2110.06864), [OCSORT](https://arxiv.org/abs/2203.14360), [DeepOCSORT](https://arxiv.org/abs/2302.11813), and [VisualRFS](https://arxiv.org/abs/2407.08872). Thanks for their wonderful works.
|