Cascaded Pyramid Network for Multi-Person Pose Estimation

被引:1154
作者
Chen, Yilun [3 ]
Wang, Zhicheng [3 ]
Peng, Yuxiang [1 ]
Zhang, Zhiqiang [2 ]
Yu, Gang [3 ]
Sun, Jian [3 ]
机构
[1] Tsinghua Univ, Beijing, Peoples R China
[2] Huazhong Univ Sci & Technol, Wuhan, Hubei, Peoples R China
[3] Megvii Inc Face, Beijing, Peoples R China
来源
2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) | 2018年
关键词
D O I
10.1109/CVPR.2018.00742
中图分类号
TP18 [人工智能理论];
学科分类号
140502 [人工智能];
摘要
The topic of multi-person pose estimation has been largely improved recently, especially with the development of convolutional neural network. However there still exist a lot of challenging cases, such as occluded keypoints, invisible keypoints and complex background, which cannot be well addressed. In this paper, we present a novel network structure called Cascaded Pyramid Network (CPN) which targets to relieve the problem from these "hard" keypoints. More specifically, our algorithm includes two stages: Global-Net and RefineNet. GlobalNet is a feature pyramid network which can successfully localize the "simple" keypoints like eyes and hands but may fail to precisely recognize the occluded or invisible keypoints. Our RefineNet tries explicitly handling the "hard" keypoints by integrating all levels of feature representations from the GlobalNet together with an online hard keypoint mining loss. In general, to address the multi-person pose estimation problem, a top-down pipeline is adopted to first generate a set of human bounding boxes based on a detector followed by our CPNfor keypoint localization in each human bounding box. Based on the proposed algorithm, we achieve state-of-art results on the COCO keypoint benchmark, with average precision at 73.0 on the COCO test-dev dataset and 72.1 on the COCO test-challenge dataset, which is a 19% relative improvement compared with 60.5from the COCO 2016 keypoint challenge. Code' and the detection results for person used will be publicly available for further research.
引用
收藏
页码:7103 / 7112
页数:10
相关论文
共 43 条
[1]
Andriluka M, 2009, PROC CVPR IEEE, P1014, DOI 10.1109/CVPRW.2009.5206754
[2]
[Anonymous], HUMAN POSE ESTIMATIO
[3]
[Anonymous], 2017, IEEE INT C COMP VIS
[4]
[Anonymous], 2017, CVPR
[5]
[Anonymous], 2015, P INT C COMP VIS ICC
[6]
[Anonymous], HUMAN POSE ESTIMATIO
[7]
[Anonymous], IEEE INT C COMP VIS
[8]
[Anonymous], 2016, LECT NOTES COMPUT SC, DOI DOI 10.1007/978-3-319-46484-8_29
[9]
[Anonymous], 2006, IEEE T COMPUT C
[10]
[Anonymous], ARXIV E PRINTS