YOLOv10-Based Real-Time Pedestrian Detection for Autonomous Vehicles

Accurate pedestrian detection is increasingly important for safety as autonomous driving technology advances. This paper presents a real-time pedestrian detection method based on YOLOv10. The technique creates an efficient real-time object detection model by enhancing the backbone network with EfficientNet and C2F-DM modules, integrating the BiFormer module in the neck network, and incorporating a multi-scale feature fusion detection head. Experimental results show that YOLOv10 can achieve efficient multi-scale pedestrian detection, even in complex backgrounds.