Overview
This dataset contains 4,284 RGB UAV images; these images constitute the annotated subset of the “Big Bird” dataset. 3,209 of the images are empty. The non-empty images include 49,990 annotations (41,337 bounding boxes and 8,653 bounding polygons) representing 102 bird species. The most common categories are “Australian pelican” (5,109 annotations), “Chinstrap penguin” (5,083 annotations), and “Chilean flamingo” (3,741 annotations).
Citation, license, and contact information
If you use this dataset, please cite:
Wilson JP, Amano T, Bregnballe T, Corregidor-Castro A, Francis R, Gallego-García D, Hodgson JC, Jones LR, Luque-Fernández CR, Marchowski D, McEvoy J, McKellar AE, Oosthuizen WC, Pfeifer C, Renner M, Sarasola JH, Sokač M, Valle R, Zbyryt A, Fuller RA. Big Bird: A global dataset of birds in drone imagery annotated to species level. Remote Sensing in Ecology and Conservation, 2026.
For questions about this data set, contact Josh Wilson.
This data set is released under the University of Queensland Permitted Re-Use with Acknowledgement License.
Data format
Annotations are provided in COCO format. The original dataset included annotations in Labelme format; those are included as well.
Metadata is available here.
