Hand pose estimation based on improved NSRM network

EURASIP Journal on Advances in Signal Processing

Research

Shiqiang Yang¹  Dexin Li²  Duo He³  Jinhua Wang⁴  Qi Li⁵ 

[1] School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, South Jinhua Road, 710048, Baiyin, Gansu Province, People’s Republic of China;School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, South Jinhua Road, 710048, Yantai, Shandong Province, People’s Republic of China;School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, South Jinhua Road, 710048, Yulin, Shaanxi Province, People’s Republic of China;School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, South Jinhua Road, 710048, Zhengzhou, Henan Province, People’s Republic of China;School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, South Jinhua Road, 710048, Zhumadian, Henan Province, People’s Republic of China;

关键词: Deep learning;    Hand pose estimation;    NSRM;    NHRNet;   

DOI  :  10.1186/s13634-023-00970-y

 received in 2022-07-15, accepted in 2023-01-03,  发布年份 2023

来源: Springer

【 摘 要 】

Hand pose estimation is the basis of dynamic gesture recognition. In vision-based hand pose estimation, the performance of hand pose estimation is affected due to the high flexibility of hand joints, local similarity and severe occlusion among hand joints. In this paper, the structural relations between hand joints are established, and the improved nonparametric structure regularization machine (NSRM) is used to achieve more accurate estimation of hand pose. Based on the NSRM network, the backbone network is replaced by the new high-resolution net proposed in this paper to improve the network performance, and then the number of parameters is decreased by reducing the input and output channels of some convolutional layers. The experiment of hand pose estimation is carried out by using public dataset, the experimental results show that the improved NSRM network has higher accuracy and faster inference speed for hand pose estimation.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

41116_2022_35_Article_IEq420.gif

41116_2022_35_Article_IEq419.gif

Fig. 6

41116_2022_35_Article_IEq417.gif

41116_2022_35_Article_IEq416.gif

41116_2022_35_Article_IEq414.gif

41116_2022_35_Article_IEq413.gif

41116_2022_35_Article_IEq412.gif

41116_2022_35_Article_IEq411.gif

41116_2022_35_Article_IEq410.gif

41116_2022_35_Article_IEq409.gif

41116_2022_35_Article_IEq408.gif

Fig. 4

41116_2022_35_Article_IEq406.gif

41116_2022_35_Article_IEq405.gif

41116_2022_35_Article_IEq404.gif

41116_2022_35_Article_IEq403.gif

12888_2022_4443_Article_IEq4.gif

12888_2022_4443_Article_IEq3.gif

12888_2022_4443_Article_IEq2.gif

12888_2022_4443_Article_IEq1.gif

41116_2022_35_Article_IEq398.gif

41116_2022_35_Article_IEq397.gif

41116_2022_35_Article_IEq396.gif

41116_2022_35_Article_IEq395.gif

41116_2022_35_Article_IEq394.gif

41116_2022_35_Article_IEq393.gif

41116_2022_35_Article_IEq383.gif

41116_2022_35_Article_IEq375.gif

41116_2022_35_Article_IEq301.gif

41116_2022_35_Article_IEq286.gif

41116_2022_35_Article_IEq281.gif

41116_2022_35_Article_IEq279.gif

41116_2022_35_Article_IEq276.gif

41116_2022_35_Article_IEq273.gif

41116_2022_35_Article_IEq270.gif

41116_2022_35_Article_IEq267.gif

41116_2022_35_Article_IEq264.gif