A deep learning approach to photovoltaic cell defect classification

The aim of this paper is to determine whether photovoltaic (PV) cells can be automatically identified as either defective or normal from electroluminescence (EL) images. This paper utilizes an experimental methodology to address the identified research problem. This paper provides evidence that deep learning (DL) can be used to distinguish between a defective and a normal PV cell. The results of this research confirm that techniques from the Computer Science discipline can be applied in photovoltaics to alleviate the tedious processes used in identifying defective PV cells from EL images.