多模态视频推理与分析竞赛
计算机视觉与模式识别
2021-08-20 v1 人工智能
摘要
在本文中，我们介绍了与 ICCV 2021 联合举办的多模态视频推理与分析竞赛（MMVRAC）研讨会。该竞赛由四个不同赛道组成，即视频问答、基于骨架的动作识别、基于鱼眼视频的动作识别以及行人重识别，这些赛道基于两个数据集：SUTD-TrafficQA 和 UAV-Human。我们总结了参赛者在该竞赛中提交的顶尖方法，并展示了他们在竞赛中取得的结果。
引用
@article{arxiv.2108.08344,
title = {The Multi-Modal Video Reasoning and Analyzing Competition},
author = {Haoran Peng and He Huang and Li Xu and Tianjiao Li and Jun Liu and Hossein Rahmani and Qiuhong Ke and Zhicheng Guo and Cong Wu and Rongchang Li and Mang Ye and Jiahao Wang and Jiaxu Zhang and Yuanzhong Liu and Tao He and Fuwei Zhang and Xianbin Liu and Tao Lin},
journal= {arXiv preprint arXiv:2108.08344},
year = {2021}
}
备注
Accepted to ICCV 2021 Workshops