Improved Yolov5n strip surface defect detection algorithm

To meet the practical application of the need of the strip surface defect detection, reduce the requirement of defect detection model for testing the deployment of hardware, network based on yolov5 algorithm, this paper proposed an improved model of strip surface defect detection. By embedding the attention mechanism in the backbone part of the network, the ability of the network to pay attention to the surface defect features of steel strips is increased, so as to improve the defect detection effect of the improved network. And by improving the loss function of the network, the training effect of the network model is improved and the convergence time of the model in training is shortened. In this paper, NEU-DET data set is used as the only experimental data set to test the improved algorithm. The experimental results show that the recall of the proposed algorithm is increased by 8.3%, the mean average precision is increased by 3.6%, the weight file size of the model remains unchanged, and the training time of the model is shortened by 44%. The experimental results show the effectiveness of the improved algorithm.