Litcius/Paper detail

Reliable and Lightweight Adaptive Convolution Network for PCB Surface Defect Detection

Lei Lei, Han‐Xiong Li, Haidong Yang

2024IEEE Transactions on Instrumentation and Measurement14 citationsDOI

Abstract

Surface defect detection is very important for the printed circuit board (PCB) to ensure their quality requirements. This paper proposes a reliable and lightweight adaptive convolution network for PCB surface defect detection. First, an automated optical inspection (AOI) for collecting PCB defects is introduced, and the formation mechanism of PCB defects is systematically analyzed. After that, lightweight adaptive convolution strategically aggregates multiple convolution kernels and simplifies model complexity through tensor decomposition. Furthermore, the confidence gate learning strategy aims to cope with dataset noise by combining collaborative learning and confidence evaluation. Complexity and convergence analyses support the theoretical basis of the method. Finally, three industrial defect datasets are used to evaluate the effectiveness. The results show the methodology has powerful feature representation, visual interpretability, and detection robustness.

Topics & Concepts

Convolution (computer science)Computer scienceElectronic engineeringSurface (topology)Materials scienceEngineeringArtificial neural networkArtificial intelligenceMathematicsGeometryIndustrial Vision Systems and Defect DetectionVehicle License Plate RecognitionImage and Object Detection Techniques
Reliable and Lightweight Adaptive Convolution Network for PCB Surface Defect Detection | Litcius