GlassEye 3-Way Unified Building Defect Dataset (640px)
The GlassEye 3-Way Unified Building Defect Dataset is a curated, multi-domain benchmark corpus designed for training computer vision models to detect structural building defects (cracks, spalling, efflorescence) across close-up façade photography and high-altitude aerial drone surveys.
Primary Repository: GlassEye GitHub
Model Checkpoint: sanjeevafk/glasseye-yolo
Image Count: 2,899 total training images + 289 held-out… See the full description on the dataset page: https://huggingface.co/datasets/sanjeevafk/glasseye-dataset.
