Recognition and Reading Method of Pointer Meters in Hard X-Ray Free-Electron Laser Facility

Due to the highly complex environment of hard X-ray free-electron laser (XFEL) facilities, the recognition of pointer meter poses significant challenges in detection precision, data reading, and real-time performance. This paper proposes an improved YOLOv8-based model with optimized neck,called Bi-CFpn-YOLO. The model is trained on the customed dataset to extract meter regions and pointers, then meter readings are calculated using the angular method after perspective transformation. Comparative experimental results show that Bi-CFpn-YOLO model can achieves the required accuracy of detection and reading while reducing computational cost in hard XFEL real-world scenarios.