AttBoT-YOLO: Beyond Efficient YOLOv5 for PCB Defect Detection

This paper proposes a lightweight defect detection algorithm called AttBoT-YOLO, aiming to solve the issues of missed detection, high false detection rate, and difficulty in identifying various types of defects in most current printed circuit board (PCB) defect detection methods. Firstly, the backbone of the network is adjusted based on the Bottleneck Transformer to enhance the feature extraction and generalization capabilities of the network. Secondly, a coordinate attention mechanism is integrated to provide better feature correlation for the model, further improving the detection accuracy. Experimental results show that the improved YOLOv5 model achieves an average precision (mAP) of 96.4% on the PCB defect dataset, which is 1.4% higher than the baseline, and achieves an accuracy of 96.1%, enabling more accurate detection of defects in industrial production of printed circuit boards.