PCB Defect Detection Method Based on Improved YOLOv11

View at publisher Summarise with VIRAT
Authors: Lianmeng Deng, Congzhe You
Cited by: 0
References: 3

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

To address defects such as short circuits, burrs, and rodent damage in printed circuit board (PCB) manufacturing—defects typically characterized by low pixel density and minimal contrast with the background—this study proposes an enhanced PCB defect detection method based on YOLOv11. Initially, the ParNet architecture is incorporated into the C3k2 module, resulting in a novel C3k2_ParNet module that improves the model’s capacity for multi-scale feature fusion and extraction, thereby enhancing its representation of complex defect patterns. Subsequently, an adaptive downsampling module (ADown) is introduced to replace certain convolutional layers; this module, in conjunction with average and max pooling operations, effectively reduces the network’s parameter count and computational complexity, leading to improved detection efficiency. Experimental evaluations indicate that, relative to the baseline YOLOv11 model, the proposed approach achieves increases of 1.6%, 1.3%, and 0.9% in precision, recall, and mean average precision (mAP), respectively. Moreover, validation on novel datasets confirms the model’s strong generalization ability, demonstrating notable adaptability and robustness.

Fields of Study & Subjects

Industrial Vision Systems and Defect Detection Advanced Neural Network Applications Infrastructure Maintenance and Monitoring

Publication details

Journal / source
IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
Pages
723–728
Publication date
28 January 2026
Language
EN