Comparative Analysis of Palmprint Detection Using Different YOLO Models

As biometric technology increasingly integrates into identity verification and security, palmprint recognition showcases significant potential due to its uniqueness, stability, and non-invasiveness. However, the accuracy and efficiency of this recognition method are heavily reliant on the precise localization and extraction of the ROI within the palmprint. This study investigates the application of the latest versions of the YOLO object detection algorithm in palmprint ROI detection and their effects on authentication system performance. The research results indicate that YOLOv9, YOLOv10, and YOLOv11 all performed excellently in terms of mAP and F1 metrics, with only minor differences among them. Yet, YOLOv9 achieved an accuracy rate of 94.44% in comprehensive evaluations, surpassing the other two generations. This result underscores the effectiveness of YOLOv9 in palmprint ROI detection and provides valuable empirical support for the optimization and application of YOLO algorithms in the field of biometric recognition.