Palmprint Classification Detection Algorithm Based on Modified CenterNet

Target detection is the most studied and widely used branch in the field of computer vision technology, mainly focusing on two issues: objection position and category information. CenterNet is one of the excellent performance methods in the one-stage target detection. In this paper, HRNet is used to replace DLA-DCN in the original network structure to detect palmprint delta and other palmprint features. The experimental results show that the average precision of the modified CenterNet in extracting palmprint delta area and other palmprint features is improved with different degree.