A Improved YOLO Model for Moving Object Detection of Slope Rockfall

A improved YOLO model for moving object detection of slope rockfall is proposed to address the problems of traditional slope rockfall detection models,including low detection accuracy under the influence of motion blur effect,poor detection of small-scale rockfall,and difficulty in distinguishing rockfall from static rock masses.To build the YOLOv5-based model,an edge enhancement module and a network structure of deepen bidirectional feature pyramid have been introduced,several small-scale probes,new convolution modules,and a dynamic rockfall filtering module have been added.The study results indicate that,compared to the YOLOv5 model,the improved model achieved improvements of 4.75%,21.89%,and 11.98% in precision,recall,and mAP respectively.The improved model has strong resistance to environmental interference.It is suitable for moving object detection of slope rockfall.