Aiming at the problems of multi-scale change and small target density, feature degradation caused by dynamic illumination and weather interference, as well as boundary box missing detection and positioning offset under overlapping occlusion, an efficient traffic image detection algorithm based on improved yolov11n is proposed. Firstly, the C3k2-DConv module is constructed by embedding dynamic convolution (DConv) in the C3k2 module of the network, which improves the sensory field expansion efficiency and dynamic feature capture capability; Secondly, the dynamic feature enhancement ASPP-LSKA module is designed to replace the backbone network spatial pyramid pooling-fast (SPPF) module with atrous spatial pyramid pooling (ASPP) module for dynamic capture of multi-scale context features and incorporates large separable kernel attention (LSKA) for more effective capture of long-range feature interactions; Additionally, the neck network is reconstructed by adding shallow feature maps to the neck layer to enhance the model’s perception ability of small targets, and combined with space to depth convolution (SPDConv) to alleviate the loss of fine-grained information in the downsampling process; Finally, Soft-NMS and DIoU-NMS are fused and replace the original NMS, which achieves retaining more bounding boxes while utilizing the distance information between the bounding boxes in order to further optimize the selection of bounding boxes and improve the bounding box accuracy. The experimental results showed that on the SODA-10M dataset, compared with the YOLOv11n model, the average detection accuracy mAP@0.5 and mAP@0.5:0.95 values increased by 4.1% and 6.7%, respectively; On the BDD100K dataset, the average detection accuracy mAP@0.5 and mAP0.5:0.95 values increased by 6.3% and 7.2%, respectively; On the VisDrone2019 dataset, the average detection accuracy mAP@0.5 and mAP@0.5:0.95 values increased by 8.5% and 6.8%, respectively. It can be seen that the proposed algorithm is effective and generalizable in traffic scene detection.