YOLOv8-LSD:A Steel Surface Defect Detection Algorithm Based On YOLOv8

In order to solve the problems of insufficient detection accuracy and high model complexity of existing steel surface defect detection algorithms, an improved model YOLOv8-LSD is designed based on the YOLOv8 model, and theLSKBlock module is introduced into the Backbone to dynamically sense the defect environment information to improve the detection accuracy. At the same time, the SCConv module is added to the neck to reduce redundant features by combining the spatial reconstruction unit and the channel reconstruction unit, which increases the accuracy of the mo del and reduces the complexity of the model. In order to further improve the performance of the model, the DRB module was introduced into the C2f module, and the model performance was further improved by combining the non-dilated large-core convolutional layer and multiple expanded small-core convolutional layers. Finally, a large number of experiments are carried out on the NEU-DET dataset on the improved model, and the results show that the map of the improved model is increased by 1. 3% compared with the baseline model, reaching 78. 3%, and the number of parameters and computations are 2. 61M and 7. 3G, which are 9% and 8% less than the baseline, respectively.