Research on Multi-target Detection Algorithm DEB-YOLO in Complex Urban Road Scenarios

This study addresses issues in object detection in complex road scenarios, such as irregular shapes, uneven lighting,especially poor image quality in low-light conditions, difficulty in recognizing small targets, and high computational cost. We developed a C2f-DCN fusion module to enhance the ability to capture useful features, adapt to irregular targets, and improve the accuracy of small target detection, also introduced the EIoU loss function to optimize bounding box prediction and model convergence speed and employed a weighted bidirectional feature pyramid network (BiFPN) to achieve rapid multi-scale feature fusion, balancing accuracy and computational efficiency. This approach improves object detection performance in complex road scenarios, and future work will focus on further optimizing the network structure to reduce computational cost and enhance detection efficiency.