Multimodal data fusion for welding defect detection using ensemble deep learning

doi:10.1016/j.autcon.2025.106694

Multimodal data fusion for welding defect detection using ensemble deep learning

2025-12-08

0

PRE

AI

DOI：10.1016/j.autcon.2025.106694

Original

Original request for help

Save

Abstract

En 中文

• Ensemble deep learning method for welding defect identification. • Surface image of weld spots are processed via a dual-input weight-sharing network. • Subnetworks are trained separately before fusion. • The proposed method achieves an overall accuracy of 91.6 % across various welding defect scenarios. • Modal-specific contribution quantification analysis is conducted.

Multimodal data fusion for welding defect detection using ensemble deep learning

Abstract

Journal

Organization