VarieD: A Multi-City Aerial Trajectory Dataset of Road-User Interactions in School-Centered Urban Environments

High-quality road-user trajectory data are essential for traffic modeling, road-safety assessment, and the analysis of interactions involving vulnerable road users (VRUs). Existing UAV-based trajectory datasets are often limited in geographic scope, facility type, or sample size. This paper introduces varieD, an aerial trajectory dataset collected across 10 municipalities in Flanders, Belgium, with a particular focus on school zones. The dataset was generated using a standardized workflow that includes privacy-preserving masking, georeferencing with ground control points, and automated road-user detection and tracking. varieD provides georeferenced trajectories for seven road-user classes across intersections, roundabouts, road segments, and kiss-and-ride facilities. The current release contains 64,534 road-user trajectories extracted from 29 h of UAV recordings collected at 28 unique observation sites across Flanders. The dataset captures variation in infrastructure characteristics, observed mobility patterns, and modal composition across the selected sites, including extensive observations of environments with high VRU activity. Beyond its value as a research dataset, varieD has supported studies on automated origin-destination flow extraction, pedestrian-vehicle interaction analysis, and surrogate safety assessment, as well as applied road-safety evaluations conducted in collaboration with municipalities. varieD provides an empirical resource for interaction-oriented transportation research, road-safety analysis, traffic modeling, and data-driven urban mobility planning for school environments.