Detection Algorithm of Bus Painting Defects Based on SAGT-YOLOV10n

The research based on the inspection of paint surface defects on passenger bus has found that the inspection has problems such as low detection accuracy, false background recognition, high model complexity and difficult identification of small target defects. However, the traditional mainstream detection algorithm cannot solve this kind of problem very well, so a passenger car painting defect detection algorithm SAGTYOLOV10n based on YOLOV10n is proposed. By adding SACCG module to YOLOV10n backbone network, TFE module is added to the model neck. SACCG module can better preserve the sparse visual features, avoid the loss of information on the 2D fractional vector, and solve the problem that the global average pooling is too coarse grained. In the process of feature fusion, In the process of feature fusion, the TFE module can better preserve the fine-grained features of the image, and improve the algorithm's ability to detect dense small targets in complex backgrounds. The experimental results on the our data set Paintdata2024 show that the average detection accuracy for all target categories has increased from 85.8% of YOLOV10n to 88.5 % of SAGT-YOLOV10n.