Safety Helmet Detection Method in Complex Industrial Scenarios

Detection of safety helmets is of great significance in the industrial field. In complex industrial scenes, there are environmental problems such as background interference, small targets, occlusion, and poor lighting. At the same time, the real-time nature of the model also needs to be considered. Among existing methods, YOLOV8 has achieved a relative balance in performance and efficiency, but its accuracy in complex scenarios has certain deficiencies. This article proposes a multiview feature pyramid MFP (Multi-granularity Feature Pyramid) network model, which uses three branches to observe objects from different angles for safety helmet detection, and combines it with YOLOV8. This model improves the detection accuracy of safety helmets in complex industrial scenes while maintaining the original advantages of YOLOV8.