Steel surface defect detection model based on YOLOv8

In order to improve the accuracy of steel surface defect detection, this study proposes a multi-directional optimized improved model based on the YOLOv8 algorithm. The C2f module in the model was improved by integrating it with DySnakeConv, leading to the innovative C2f_DySnakeConv module, which enhances the model's ability to capture local features. Additionally, the SPDConv module was added to the model to enhance its capability to process data with specific geometric structures and improve its generalization ability. Furthermore, the neck part of the model was modified by incorporating the Zoom_cat module, which improves the model's ability to detect and segment objects of varying sizes. The ScalSeq module was introduced to fuse features at different scales, enabling the model to better capture targets of different sizes and shapes. Experimental results show that the improved model achieves an average precision (mAP) of 83.6% in steel surface defect detection tasks, a significant improvement of 6.2% compared to the original YOLOv8 model. The improved model meets the requirements of steel surface defect detection in industrial applications.