Research on Fluorescent Magnetic Particle Inspection of Industrial Part Defects Based on Improved YOLOv5

To meet rising quality inspection demands in industrial automation, this study presents an enhanced YOLOv5 model for detecting minor surface defects in industrial parts through fluorescent magnetic particle inspection (FMPI). Traditional FMPI relies on inefficient and subjective manual checks. Our improved model, featuring a Channel-Attention module and optimized for a custom dataset, significantly boosts detection precision and speed. Experiments confirm its superior performance over conventional methods, offering a robust solution for automated non-destructive testing.