Automating PV Module Health Inspections with Machine Learning.
Defect detection, automation, machine learning, computer vision, image analysis, diagnostics, electroluminescence imaging
Our mission is to help PV companies get actionable information from electroluminescence images — lightning fast and at a fraction of the price of other analysis techniques.
As thousands of electroluminescence (EL) images are collected each week, our intensive automated image analysis solution can be used to help diagnose a module's health for those all across the photovoltaic supply chain.
17,094
17,094
Our models are trained on a pixel-level annotated dataset, comprised of 17,094 cells.
9
9
Success and research on 9 different defect categories, including cracks, contact defects, and interconnect defects.
95.4%
95.4%
Pixel segmentation accuracy of 95.4%.
Based out of Orlando, FL
