Overview
This dataset contains 37,075 images from the WWF-UK/UCL Biome Health Project site in the Maasai Mara, Kenya. These images form the training, test, and validation sets for the Maasai Mara Classifier model. This dataset was created by labeling one image per five-minute period across 176 camera sites.
Labels are provided for 100 categories, most of which are species, but some of which are species groups (for example, the ‘shoat’ category is sheep and goats combined). Labels include wild mammals, wild birds, and domestic mammals. Blank images and images containing humans have been removed.
Citation, license, and contact information
If you use this dataset, please cite:
Connolly E, Pringle HA, Pantazis O, Ferreira GB, Madsen EK, Ingram DJ, Bains T, Brostow GJ, Carroll S, Cronshaw G, De Ornellas P, Di Minin E, Ewers RM, Gichangi K, Mac Aodha O, Mulama M, Njuguna M, Pattullo L, Pickering A, Rabeau A, Rowcliffe M, Spooner F, Thomas L, Wato Y, Woodhouse E, Collen B, Mace GM, Jones KE. Sustainable cattle management by communities supports African wildlife. bioRxiv. 2025 Oct 10:2025-10.
For questions about this data set, contact Kate Jones at University College London.
This data set is released under the Community Data License Agreement (permissive variant).
Data format
Annotations (including species tags and unique location identifiers) are provided in COCO Camera Traps format.
For information about mapping this dataset’s categories to a common taxonomy, see this page.
Downloading the data
Metadata is available here.
