Automotive Gear Defect Detection Method based on Yolov8 Algorithm

For the problem of low efficiency and accuracy of automotive gear defect detection by manual visual inspection method, a detection method of automotive gear endface defects based on Yolov8s algorithm is proposed. The original Yolov8s model is improved in this paper considering the problem of smaller defects in automotive gears. Firstly, the CBAM attention mechanism was introduced in Backbone to help the model pay more attention to information related to gear defects while suppressing useless information in the data. It better captures the key features of gear defects. Secondly, a small target detection layer is added to the network structure to solve the problem of small target information loss due to the large downsampling multiplier of Yolov8s. It better extracts the features of small targets. For the problem that some defective kinds of samples in the dataset collected from real industrial scenes are too small, this paper uses the Mosaic data augmentation method, which is used to expand the number of samples and solve the problem of inadequate training to some extent. The improved Yolov8s was compared with the original Yolov8s model for experiments. The results show that the improved Yolov8s model improves both Precision and Recall over the original Yolov8s model. Meanwhile, the improved Yolov8 algorithm is used for automotive gear defect detection in a real scenario detection task, which effectively reduces the miss detection rate and shows good performance. It meets the inspection needs of real industrial scenarios, and is advanced and practical in automotive gear defect detection.