2018 WIDER 人脸与行人挑战赛：方法与结果
计算机视觉与模式识别
2019-02-20 v1
摘要
本文对 2018 年 WIDER 人脸与行人挑战赛进行综述。该挑战赛聚焦于人体面部与身体的精确定位，以及身份信息的准确关联。其包含三个赛道：（i）WIDER Face，旨在征集新方法以推进人脸检测的最先进水平（SOTA）；（ii）WIDER Pedestrian，旨在寻找有效且高效的方法以解决无约束环境中行人检测问题；（iii）WIDER Person Search，提出了在 192 部电影中检索行人的激动人心的挑战。总计 73 支队伍向各赛道提交了有效方案。我们总结了三个赛道的获胜方案，并就这些主题中的开放问题与潜在研究方向展开讨论。
引用
@article{arxiv.1902.06854,
title = {WIDER Face and Pedestrian Challenge 2018: Methods and Results},
author = {Chen Change Loy and Dahua Lin and Wanli Ouyang and Yuanjun Xiong and Shuo Yang and Qingqiu Huang and Dongzhan Zhou and Wei Xia and Quanquan Li and Ping Luo and Junjie Yan and Jianfeng Wang and Zuoxin Li and Ye Yuan and Boxun Li and Shuai Shao and Gang Yu and Fangyun Wei and Xiang Ming and Dong Chen and Shifeng Zhang and Cheng Chi and Zhen Lei and Stan Z. Li and Hongkai Zhang and Bingpeng Ma and Hong Chang and Shiguang Shan and Xilin Chen and Wu Liu and Boyan Zhou and Huaxiong Li and Peng Cheng and Tao Mei and Artem Kukharenko and Artem Vasenin and Nikolay Sergievskiy and Hua Yang and Liangqi Li and Qiling Xu and Yuan Hong and Lin Chen and Mingjun Sun and Yirong Mao and Shiying Luo and Yongjun Li and Ruiping Wang and Qiaokang Xie and Ziyang Wu and Lei Lu and Yiheng Liu and Wengang Zhou},
journal= {arXiv preprint arXiv:1902.06854},
year = {2019}
}
备注
Report of ECCV 2018 workshop: WIDER Face and Pedestrian Challenge