MOT20: A benchmark for multi object tracking in crowded scenes
Paper • 2003.09003 • Published
Error code: RowsPostProcessingError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
MOT20 是 MOTChallenge 多目标跟踪基准的 2020 年版,专门面向极高人群密度场景,用于推进遮挡条件下的行人跟踪研究。
MOT20-01 ~ MOT20-08。| 文件 | 说明 |
|---|---|
MOT20.zip |
图像序列 |
MOT20Labels.zip |
训练集标注 |
{MOT20 ROOT}
├── train/
│ └── MOT20-01/
│ ├── img1/ # 图像序列
│ ├── det/det.txt # 公开检测结果
│ └── gt/gt.txt # 标注
└── test/
└── MOT20-06/
├── img1/
└── det/det.txt
gt.txt)
每行 10 个字段:
<frame>, <id>, <bb_left>, <bb_top>, <bb_width>, <bb_height>, <conf>, <class>, <visibility>, <ignore>
frame、id 从 1 开始;conf = 0 表示该条目不参与评测。x, y, z 可填 -1。官方评测请通过 MOTChallenge 平台或 CodaBench 提交,工具推荐 TrackEval。
@article{dendorfer2020mot20,
title={MOT20: A Benchmark for Multi Object Tracking in Crowded Scenes},
author={Dendorfer, Patrick and Rezatofighi, Hamid and Milan, Anton and Shi, Javen and Cremers, Daniel and Reid, Ian and Roth, Stefan and Schindler, Konrad and Leal-Taix{\'e}, Laura},
journal={arXiv preprint arXiv:2003.09003},
year={2020}
}
CC BY-NC-SA 3.0
MOTChallenge 数据集的官方声明为 Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License:
协议全文:https://creativecommons.org/licenses/by-nc-sa/3.0/
MOT20 包含真实公共场所的行人影像。使用者在训练与部署模型时需考虑隐私、监控与公平性影响。已知局限:测试集标注不公开;场景不覆盖所有国家、相机类型、光照与人群构成;模型在真实监控或安全关键场景部署前应谨慎评估。
权威访问条款请以 https://motchallenge.net/data/MOT20/ 为准,本镜像不授予超出原始协议的额外权利。