MSPS: Monocrystalline Silicon Photovoltaic Cell Defect Segmentation Dataset

MSPS is an expert-annotated dataset for pixel-level defect segmentation of monocrystalline silicon photovoltaic cells in electroluminescence (EL) images. It contains 1,500 production-line images at 640 × 590 pixels, including 1,000 training images and 500 test images. The annotations cover defects such as busbar breaks, black spots, and cracks. Each image was independently annotated by two experts, with disagreements adjudicated by a third expert. The dataset supports research on photovoltaic-cell defect segmentation, industrial visual inspection, and severe foreground–background class imbalance. The associated paper is “DACN: Dynamic Adjustment Contrast Network for Photovoltaic Cell Defect Segmentation.” Data organization and usage information are available at https://github.com/wch313/MSPS-Dataset.