The recruitment of public volunteers (“citizen scientists”) to assist with research has become a popular method of processing big data in the Information Age. Snapshot Safari, an ecological monitoring network designed for long-term observation of endangered wildlife populations throughout eastern and southern Africa, has attracted thousands of volunteers to help classify the species, count, and behavior of animals caught on camera in various African parks and reserves. Camera trap grids regularly spaced every 5km2 at 28 national parks and game reserves are currently running. The consensus classifications and raw imagery provide an unparalleled opportunity to investigate multi-species dynamics in multiple ecosystems and a valuable resource for machine-learning and computer-vision research. The strategy of incorporating machine learning may provide the key to maximizing volunteer effort and quickly producing usable data for research and conservation management purposes. Data from this group's original project, Snapshot Serengeti, has been used to train several machine learning algorithms and is being developed to recognize individual animals within populations as well. Snapshot Safari has teamed with Zooniverse, Microsoft, and the University of Wyoming AI Lab to create a massive labeled image library ("LILA") that contains the classified images and is freely available for use in training other algorithms or for use by ecologists. The group's machine-learning algorithms continue to be retrained and refined.
This research was featured on the MSI website in:
- February 2018: New Citizen-Science Project, Snapshot Safari, Launches
- July 2015: Wildlife in the Serengeti