A Lightweight Laser Chip Defect Detection Algorithm Based on Improved YOLOv7-Tiny

[Purposes] Catastrophic Optical Damage (COD) is a major limiting factor for the reliability and lifespan of high-power semiconductor lasers, making effective defect detection crucial for optimizing the manufacturing processes and structural designs of laser chips. In this study, a lightweight laser chip defect detection algorithm based on an improved YOLOv7-Tiny is proposed, aiming at addressing the high computational and parameter demands of deep learning applications in defect detection. [Methods] By employing a lightweight convolutional neural network as the feature extraction backbone and integrating multi-branch reparameterized convolution blocks, this algorithm not only significantly reduces resource consumption but also enhances feature representation capabilities. Additionally, the introduced coordinate attention mechanism improves the precision of defect localization. Pruning experiments and model deployment are conducted to further verify the algorithm practicality. [Findings] Experimental results on the electroluminescence dataset demonstrate that this method can accurately detect chip defects with lower parameter and computational costs, showing excellent performance.