Steel surface defect detection algorithm based on EBC-YOLOv8

Aiming at the problems of insufficient recognition accuracy and insufficient feature fusion in steel surface defect detection, this paper proposes an EBC-YOLOv8 steel surface defect detection algorithm. On the basis of YOLOv8, the algorithm introduces an efficient feature extraction attention module to enhance the expression ability of key features, reconstructs the neck structure of the model, and uses Bidirectional Feature Pyramid Network (BiFPN) to achieve effective fusion of multi-level features. At the same time, the coordinate attention mechanism is introduced to further improve the model's ability to perceive the target spatial information. Experimental results show that compared with the original YOLOv8 model, the EBC-YOLOv8 model has the recall rate increased by 6.5% and the mean average precision increased by 2.1% in the steel surface defect detection task, while the model parameter number decreases by 0.1M, which effectively improves the detection performance and overall recognition ability of small target defects.