UAV RGB dataset with instance-level annotations for Cortaderia selloana detection

This dataset contains 758 high-resolution UAV RGB images acquired in Galicia (Spain) and northern Portugal for the detection of Cortaderia selloana. The images were collected under varied environmental and illumination conditions using low-cost multirotor UAV platforms.

The dataset includes 29,829 instance-level bounding-box annotations, where each annotation corresponds to a single plant and is focused on the tussock structure. Annotations are provided in YOLO format, with normalized coordinates, and as a Label Studio JSON file containing the complete annotation set.

The dataset captures variability in plant density, spatial distribution, and environmental context, including both sparse and densely clustered scenarios. Most annotated objects occupy a small fraction of the image area.

The dataset is intended for use in object detection tasks and supports reproducible research in UAV-based ecological monitoring.