Annotated datasets optimized for machine learning and computer vision applications in marine optics.
These datasets have been carefully curated and annotated for training, validation, and testing of machine learning models. All annotations follow standardized formats (COCO, YOLO, etc.) and include comprehensive metadata for reproducible research.
Bounding box annotations for detecting fish, corals, structures, and equipment in underwater imagery.
Species and habitat labels for image classification, including CoralNet exports and taxonomic annotations.
Pixel-level masks and polygons for coral colonies, benthic cover, and detailed habitat mapping.
Curated train/validation/test splits with standardized formats for immediate model development.
Standardized evaluation datasets for comparing model performance across marine optics tasks.
Web-based tools and workflows for creating and validating annotations on marine imagery.