Human Pose Estimation: Multi-stage Network Based on HRNet. Issue 1 (1st December 2022)

• Record Type: Journal Article
• Authors: Ji, Xiaodong; Yang, Qiaoning; Yang, Xiuhui; Zheng, Jiahao; Gong, Mengyan
• Is Part Of: Journal of physics. Volume 2400 Issue 1(2022)
• Publication Date: 2022-12-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2400/1/012034
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Multi-stage network uses stacked networks to enhance the feature extraction capability, and can gradually refine the keypoints with the information of previous stages' output. Obviously, multi-stage networks are more suitable for human pose estimation. However, most current multi-stage networks use a codec structure as the backbone in which downsample will cause information loss. HRNet maintains high-resolution features to supply the information which is lost in down-sampling stage. In this regard, we propose a novel two-stage network with HRNet as the backbone and stacked codec structure. HRNet has more efficient feature extraction capability, and the stacked codec network can utilize the multi-scale features generated by HRNet more effectively. This method obtains a 1.2AP improvement compared to HRNet and a significant improvement compared to other two-stage networks.