DroneWaste targets automated identification of waste materials from drone-derived orthomosaics, supporting environmental inspection workflows. Each visible waste instance is annotated with a segmentation mask, bounding box, and category mapped to the European Waste Code (EWC).
BaleUAVision provides large-scale, field-diverse aerial imagery of hay bale fields with human annotations and multiple annotation formats (COCO/YOLO/CSV/JSON/masks), plus orthomosaics to support mapping and simulation workflows.
A benchmark dataset built from real 3PL operations. It provides privacy-preserving matrices (no raw addresses) and richly structured constraints to test VRP solvers and learning-based routing.
CoFly-WeedDB includes expert agronomist annotations for three common weed species (Johnson grass, field bindweed, purslane) and is suitable for training segmentation models in precision agriculture.