UCF-EL-GDI

UCF-EL-GDI is a collection of synthetic photovoltaic electroluminescence (EL) images generated with Generative Defect Isolation (GDI) from the public UCF-EL-Defect dataset.

GDI uses the original segmentation annotations to remove selected defects with neural inpainting, producing isolated single-defect samples and No_Defect samples. Full methodology, generation instructions, and implementation details are available in the project repository.

Contents

Images are organized by generated class:

Contact_BeltMarks/
Contact_Corrosion/
Contact_FrontGridInterruption/
Contact_NearSolderPad/
Crack_Closed/
Crack_Isolated/
Crack_Resistive/
Interconnect_BrightSpot/
Interconnect_Disconnected/
Interconnect_HighlyResistive/
No_Defect/

Links

• Paper: A Generative Approach for Improving Multi-Label Defect Classification in Photovoltaic Modules
• Code: mueez-overflow/generative-defect-isolation
• Original dataset: ucf-photovoltaics/UCF-EL-Defect

Citation

If you use GDI, the generated synthetic data, or any part of this work, please cite:

@article{mueez2026gdi,
  title={A Generative Approach for Improving Multi-Label Defect Classification in Photovoltaic Modules},
  author={Mueez, Abdul and Rawat, Yogesh S. and Vyas, Shruti},
  journal={Solar Energy},
  year={2026},
  doi={10.1016/j.solener.2026.114943}
}