摘要：

Mobile inspections conducted by intelligent tunnel robots are instrumental in broadening the inspection reach, economizing on expenditures, and augmenting operational efficiency of inspections. Despite differences from fixed surveillance, mobile-captured traffic videos have complex backgrounds device conditions that interfere with accurate event identification, warranting more research. This paper proposes an improved algorithm based YOLOv9 DeepSORT for detection edge computing mobile using robot. The enhancements comprise integration Temporal Shift Module to boost temporal feature recognition establishment logical rules identifying diverse incidents video imagery. Experimental results show our fused achieves a 93.25% accuracy rate, improvement 1.75% over baseline. is also applicable vehicles, drones, autonomous effectively enhancing events improving safety.

DOI：10.3390/bdcc8110147