Improved YOLOv7-Based Algorithm for Detecting Defect on Steel Surface

Abstract Steel in the production and processing process is easily affected by raw materials and processing environment, resulting in surface defects, affecting its corrosion resistance and service life. In addition, the defect detection model based on deep learning is relatively large, and it is difficult to meet the demand for real-time steel surface defect detection tasks. To address the above problems, this paper proposes an improved model YOLOv7-Lig based on the YOLOv7-tiny model. by introducing the CBAM attention mechanism, as well as using the Wise-IoU loss function to replace the original CIoU loss function, the network model is enhanced to pay attention to the key features of the input data, so that it can accurately capture the true boundary of the target, effectively reduces the problem of inadequate feature extraction. The CSP-Faster module is introduced in the BACKBONE, which reduces the number of parameters of the net work model and helps the network model to be deployed in resource-constrained environments. Then the CARAFE upsampling operator is introduced at the head, which effectively aggregates the context information and improves the accuracy of the upsampling. The YOLOv7-Lig model achieves a mAP of 85.1% and a detection speed of 82FPS, which proves that the model can improve the accuracy and detection speed of the model while reducing the number of parameters.