CTL-YOLO: A Surface Defect Detection Algorithm for Lightweight Hot-Rolled Strip Steel Under Complex Backgrounds

Currently, in the domain of surface defect detection on hot-rolled strip steel, detecting small-target defects under complex background conditions and effectively balancing computational efficiency with detection accuracy presents a significant challenge. This study proposes CTL-YOLO based on YOLO11, aimed at efficiently and accurately detecting blemishes on the surface of hot-rolled strip steel in industrial applications. Firstly, the CGRCCFPN feature integration network is proposed to achieve multi-scale global feature fusion while preserving detailed information. Secondly, the TVADH Detection Head is proposed to identify defects under complex textured backgrounds. Finally, the LAMP algorithm is used to further compress the network. The proposed algorithm demonstrates excellent performance on the public dataset NEU-DET, achieving a mAP50 of 77.6%, representing a 3.2 percentage point enhancement compared to the baseline algorithm. The GFLOPs is reduced to 2.0, a 68.3% decrease compared to the baseline, and the Params are reduced to 0.40, showing an 84.5% reduction. Additionally, it exhibits strong generalization capabilities on the public dataset GC10-DET. The algorithm can effectively improve detection accuracy while maintaining a lightweight design.