Computer vision-based algorithm for precise defect detection and classification in photovoltaic modules.

Area of Science:

• Renewable Energy
• Artificial Intelligence
• Computer Vision

Background:

• Solar power is vital globally, necessitating high-quality photovoltaic modules.
• Automated defect detection is crucial for maintaining solar energy system efficiency.
• Identifying minor defects in complex environments remains a challenge.

Purpose of the Study:

• To develop an advanced automated system for detecting minor defects in photovoltaic modules.
• To enhance the precision and speed of defect identification in solar panel inspection.
• To improve the overall quality and operational reliability of solar power systems.

Main Methods:

• Utilized a progressive annotation approach for precise defect sample labeling.
• Employed computer vision techniques for accurate segmentation of modules and defects.
• Deployed a transfer learning model, specifically Mask-Region Convolutional Neural Network (Mask R-CNN), for defect classification.

Main Results:

• Achieved high accuracy (98.7%) and recall (0.913) in defect classification.
• Demonstrated a fast detection speed of 280.69 frames per second.
• Recorded a low inference time of 3.53 milliseconds for defect identification.

Conclusions:

• The developed computer vision algorithm significantly improves automated detection of minor photovoltaic module defects.
• The Mask R-CNN model offers a highly accurate and efficient solution for solar panel quality control.
• This advancement is critical for ensuring the reliability and performance of solar energy infrastructure.