AthleticsPose
收藏资源简介:
AthleticsPose数据集是首个公开的3D人体姿态数据集,包含23名运动员在田径场上进行的各种田径项目的“真实”动作。该数据集使用8个同步摄像头进行采集,包含大约500,000个真实的田径动作帧。数据集的创建过程经过了严格的校准和同步,确保了数据的准确性和一致性。该数据集旨在解决现有数据集缺乏真实体育动作数据的问题,并为研究社区提供有价值的资源,以便更好地理解和分析运动中的动态特征。
The AthleticsPose dataset is the first publicly available 3D human pose dataset. It encompasses "authentic" motions of 23 athletes performing various track and field events on athletic fields. Collected with 8 synchronized cameras, the dataset contains approximately 500,000 frames of real track and field motion data. Its development process has undergone rigorous calibration and synchronization to ensure data accuracy and consistency. This dataset aims to address the scarcity of real-world sports motion data in existing datasets, providing a valuable resource for the research community to better understand and analyze dynamic characteristics during athletic movements.
AthleticsPose 数据集概述
基本信息
- 数据集名称:AthleticsPose: Authentic Sports Motion Dataset on Athletic Field and Evaluation of Monocular 3D Pose Estimation Ability
- 状态:开发中(Work in Progress)
- 发布日期:2025-07-18(初始版本)
- 许可证:CC BY-NC-SA 4.0
数据集内容
- 类型:运动场景下的3D姿态估计数据集
- 特点:
- 专注于田径运动场景
- 包含真实运动动作数据
- 提供单目3D姿态估计评估框架
相关论文
- 标题:AthleticsPose: Authentic Sports Motion Dataset on Athletic Field and Evaluation of Monocular 3D Pose Estimation Ability
- 作者:Tomohiro Suzuki, Ryota Tanaka, Calvin Yeung, Keisuke Fujii
- arXiv链接:https://arxiv.org/abs/2507.12905
- 年份:2025
开发计划
- 待完成事项:
- 添加数据集文件及下载说明
- 完善README.md文件
- 实现任意视频输入推理功能
- 提供预训练模型权重
- 添加数据集结构和运行方法文档
安装方法
bash git clone https://github.com/your-username/AthleticsPose.git cd AthleticsPose uv sync
引用格式
bibtex @misc{suzuki2025athleticsposeauthenticsportsmotion, title={AthleticsPose: Authentic Sports Motion Dataset on Athletic Field and Evaluation of Monocular 3D Pose Estimation Ability}, author={Tomohiro Suzuki and Ryota Tanaka and Calvin Yeung and Keisuke Fujii}, year={2025}, eprint={2507.12905}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2507.12905}, }
联系方式
- 通过GitHub提交issue或联系作者

- 1AthleticsPose: Authentic Sports Motion Dataset on Athletic Field and Evaluation of Monocular 3D Pose Estimation Ability名古屋大学 · 2025年



