Advancing Photovoltaic Cells Defect Detection with Deep Learning

Strategic Insights -

Researchers at Hohai University have published a study on using deep learning techniques to enhance defect detection in electroluminescence (EL) images of photovoltaic (PV) cells. The study aimed to explore the potential of multiple state-of-the-art object detectors in detecting defects such as cracks, finger interruptions, black cores, and thick line defects. The research concluded that the proposed YOLOv9 GELAN-e with PGI and GELAN architectures achieves promising results of 94.30% mean Average Precision (mAP) at 0.5. The study also demonstrated the network's ability to perform well on unseen data obtained in unknown imaging conditions, making it a promising solution for practical applications.

Key Takeaways:

  • The study utilized deep learning techniques to enhance defect detection in EL images of PV cells.
  • The researchers explored the potential of multiple state-of-the-art object detectors, including Detection Transformer, EfficientDet, FasterRCNN, YOLOv7, YOLOv8, and YOLOv9.
  • The proposed YOLOv9 GELAN-e with PGI and GELAN architectures achieved promising results of 94.30% mAP@0.5.
  • The study demonstrated the network's ability to perform well on unseen data obtained in unknown imaging conditions.
  • The research aimed to provide insights into real challenges at the practical level and elucidate possible solutions.
  • The study was conducted by Jianbo Bai and his team at Hohai University.
  • The research was funded by the National Key Research & Development Program of China and the Fundamental Research Funds for the Central Universities Program of China.

Statistics:

  • 94.30% mean Average Precision (mAP) at 0.5 achieved by the proposed YOLOv9 GELAN-e with PGI and GELAN architectures.
  • 6 state-of-the-art object detectors explored in the study, including Detection Transformer, EfficientDet, FasterRCNN, YOLOv7, YOLOv8, and YOLOv9.
  • 4 types of defects detected by the network, including cracks, finger interruptions, black cores, and thick line defects.

Sources:

  • Advancing Photovoltaic Cells Defect Detection In Electroluminescence Images Through Exploring Multiple Object Detectors. Solar Energy Materials and Solar Cells, 2025;292.
  • Solar Energy Materials and Solar Cells. Elsevier. www.journals.elsevier.com/solar-energy-materials-and-solar-cells/
  • NewsRx. Reports from Hohai University Provide New Insights into Photovoltaic Cells (Advancing Photovoltaic Cells Defect Detection In Electroluminescence Images Through Exploring Multiple Object Detectors). Electronics Newsweekly. October 21, 2025; p 1412.