data.zip

This dataset consists of electroluminescence (EL) images of solar photovoltaic (PV) modules, collected for the purpose of defect detection, classification, and non-destructive evaluation. Electroluminescence imaging is a widely used technique in PV quality inspection, as it enables visualization of micro-cracks, broken fingers, inactive areas, shunts, and other structural defects that are invisible under standard optical inspection. The dataset includes high-resolution grayscale images of PV modules under different conditions :Normal modules without visible defects. Defective modules containing cracks, broken busbars, finger interruptions, dark areas, and other anomalies. Each image is annotated with corresponding defect categories and metadata (e.g., defect type, defect location, image resolution). This makes the dataset suitable for multiple computer vision tasks, including: Binary classification (defective vs. non-defective modules)Multi-class classification (different types of PV defects)Object detection (bounding boxes of defects)Semantic segmentation (pixel-level defect localization)Anomaly detection (unsupervised approaches)The dataset is particularly valuable for researchers working in renewable energy, computer vision, signal processing, and intelligent fault diagnosis. It supports the development of automated inspection systems, deep learning models, and non-destructive testing (NDT) methods for PV manufacturing and field monitoring.