Steel surface defect detection algorithm based on Tcsc-Yolov8

To address the issues of insufficient feature extraction and inadequate multi-scale information fusion in steel surface defect detection, this paper proposes an improved YOLOv8s-based algorithm named TCSC-YOLOv8. Building upon the YOLOv8s framework, the algorithm enhances the C2f module in the backbone network by introducing a triple attention mechanism to construct a TA_C2f module, thereby strengthening the model's ability to emphasize key features and suppress redundant ones. In the neck structure, a lightweight CARAFE upsampling operator is incorporated to achieve more efficient feature reconstruction and multi-scale fusion. Furthermore, to enhance the model's responsiveness to salient defect regions, a Squeeze-and-Excitation (SE) attention mechanism is introduced before the detection head, significantly improving detection capability. Experimental results show that compared to the original YOLOv8s model, the improved TCSC-YOLOv8 model achieves a 6.7% increase in precision and a 3.1% improvement in mAP.