A Defect Detection Algorithm for Optoelectronic Detectors Utilizing GLV-YOLO

Photodetectors are indispensable in a multitude of applications, with the detection of surface defects serving as a cornerstone for their production and advancement. To meet the demands of real-time and accurate defect detection, this paper introduces an optimization algorithm based on the GLV-YOLO model tailored for photodetector defect detection in manufacturing settings. The algorithm achieves a reduction in the model complexity and parameter count by incorporating the GhostC3_MSF module. Additionally, it enhances feature extraction capabilities with the integration of the LSKNet_3 attention mechanism. Furthermore, it improves generalization performance through the utilization of the WIoU loss function, which minimizes geometric penalties. The experimental results showed that the proposed algorithm achieved 98.9% accuracy, with 2.1 million parameters and a computational cost of 7.0 GFLOPs. Compared to other methods, our approach outperforms them in both performance and efficiency, fulfilling the real-time and precise defect detection needs of photodetectors.