Improved human posture recognition for HRNet

In order to ensure the further safety of workers in circuit secondary maintenance, this paper proposes a new network RSHRNet to improve the accuracy of model detection by using HRNet s the backbone network for network optimization. First of all, faced with the problem of poor fitting effect of deep neural network or data loss in convolution operation, residual module is added to HRNet to form RSHRNet network to prevent data loss. In order to ensure that the data will not be overfitted or underfitted, BN layer is added for normalization processing. The residuals are stacked 1:1 between those with and without jump links to form BigRes modules. The experimental results show that compared with the benchmark model HRNet method, the improved network model significantly improves the detection accuracy of human pose estimation, the average accuracy of COCO in the public data set is up to85.8%, and the AP in MPII in the public data set is up to90.8%.