CASIA OpenIR  > 模式识别国家重点实验室  > 图像与视频分析
Keypoint Context Aggregation For Human Pose Estimation
Wenzhu, Wu1,2; Weining, Wang1,2; Longteng, Guo1,2; Jing, Liu1,2
2021-09
Conference NameInternational Conference on Image and Graphics
Conference Date2021-12-26
Conference Place中国海口
Abstract

Human pose estimation has drawn much attention recently, but it remains challenging due to the deformation of human joints, the occlusion between limbs, etc. And more discriminative feature representations will bring more accurate prediction results. In this paper, we explore the importance of aggregating keypoint contextual information to strengthen the feature map representations in human pose estimation. Motivated by the fact that each keypoint is characterized by its relative contextual keypoints, we devise a simple yet effective approach, namely Keypoint Context Aggregation Module, that aggregates informative keypoint contexts for better keypoint localization. Specifically, first we obtain a rough localization result, which can be considered as soft keypoint areas. Based on these soft areas, keypoint contexts are purposefully aggregated for feature representation strengthening. Experiments show that the proposed Keypoint Context Aggregation Module can be used in various backbones to boost the performance and our best model achieves a state-of-the-art of 75.8% AP on MSCOCO test-dev split.

Indexed ByEI
Language英语
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/48591
Collection模式识别国家重点实验室_图像与视频分析
Corresponding AuthorJing, Liu
Affiliation1.National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences
2.School of Artificial Intelligence, University of Chinese Academy of Sciences
First Author AffilicationChinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
Corresponding Author AffilicationChinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China
Recommended Citation
GB/T 7714
Wenzhu, Wu,Weining, Wang,Longteng, Guo,et al. Keypoint Context Aggregation For Human Pose Estimation[C],2021.
Files in This Item: Download All
File Name/Size DocType Version Access License
Wu2021_KeypointConte(1947KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Wenzhu, Wu]'s Articles
[Weining, Wang]'s Articles
[Longteng, Guo]'s Articles
Baidu academic
Similar articles in Baidu academic
[Wenzhu, Wu]'s Articles
[Weining, Wang]'s Articles
[Longteng, Guo]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Wenzhu, Wu]'s Articles
[Weining, Wang]'s Articles
[Longteng, Guo]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: Wu2021_KeypointContextAggregationForHumanPoseEstimation.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.