To address the challenges of detecting small traffic signs caused by their small size, complex backgrounds, and varying lighting and weather conditions, this paper proposes an attention-based small object detection algorithm. First, a Dual Attention Hybrid Convolution (DAHC) module is designed to jointly extract target features in the spatial and channel dimensions, effectively enhancing the feature representation of small targets. Second, an Efficient High-level Screening-feature Fusion Pyramid Network (E-HSFPN) is constructed to replace the conventional feature fusion structure, and a P2 detection layer is introduced to improve multi-scale feature fusion and further enhance small object detection accuracy. Finally, a Cascaded Multi-Feature Interaction (CMFI) module is proposed to alleviate the loss of fine-grained information in small targets and strengthen the collaborative perception of local details and global contextual information. Experimental results on the TT100K dataset demonstrate that the proposed method achieves improvements of 14.1% in mAP@0.5 and 11.3% in mAP@0.5:0.95 compared with the YOLOv11n model, validating the effectiveness of the proposed algorithm for traffic sign small object detection.