A Generative Approach for Improving Multi-Label Defect Classification in Photovoltaic Modules

2026-08-12

AI Summary

This work addresses the challenges of feature ambiguity caused by co-occurring multi-label defects and the scarcity of rare defect samples in photovoltaic electroluminescence imaging. To tackle these issues, the authors propose a Generative Defect Isolation (GDI) framework, which introduces generative inpainting for multi-label defect disentanglement. GDI leverages the LaMa image inpainting model enhanced with fast Fourier convolutions to precisely remove specific defects from multi-defect images, thereby synthesizing high-quality single-defect samples for training. Experimental results demonstrate that GDI substantially mitigates learning ambiguity induced by defect co-occurrence, achieving up to a 63.6% improvement in F1 score for rare defect classes under low-data regimes and reducing misclassification of co-occurring defects by 26%, establishing a new state-of-the-art benchmark in this domain.

Abstract

This paper addresses the challenge of multi-label defect classification in electroluminescence (EL) images of photovoltaic (PV) cells. Training models on images where multiple defects co-occur creates learning ambiguity, making it difficult to disentangle visual features for specific defect types, a problem compounded by the scarcity of examples for individual classes. To tackle this, we introduce Generative Defect Isolation (GDI), utilizing the LaMa inpainting model with Fast Fourier Convolutions to remove selected defects and generate realistic, single-defect training samples. Extensive experiments on Vision Transformer (ViT-S, ViT-L) and EfficientNetV2-L architectures demonstrate that GDI significantly outperforms baselines. The performance gains are most pronounced in low-data scenarios; class-wise analysis shows substantial improvements, boosting the F1-Score for rare defect classes by up to 63.6%. Furthermore, GDI effectively resolves learning ambiguity from co-occurring defects, yielding a 26% reduction in such co-occurring classification errors. Our work establishes GDI as an effective method for maximizing the value of existing segmentation datasets and sets a new performance benchmark for multi-label classification in this domain.

Problem

• multi-label defect classification
• photovoltaic modules
• learning ambiguity
• co-occurring defects
• data scarcity

Innovation

• Generative Defect Isolation
• multi-label classification
• defect disentanglement
• image inpainting
• photovoltaic defect detection