Citizen-science projects include thousands of people who sort through huge datasets. Now these human volunteers have a partner - computers with sophisticated image-recognition capabilities. A recent study led by University of Minnesota researchers used machine-learning techniques to “teach” computers to identify different types of animals based on camera-trap photos from several datasets. The article was the cover story for the January 2019 issue of the British Ecological Society’s journal Methods in Ecology and Evolution. A story about the project appears on the College of Science and Engineering website: Citizen Science Projects Have a Surprising New Partner - the Computer.
One of the datasets used for this study is Snapshot Serengeti, for which MSI provided image processing and data storage. Related projects for which MSI is providing support are Snapshot Safari and Eyes on the Wild.
MSI has installed software and created tutorials to assist users getting started with machine-learning techniques, and is continuing to test new environments and hardware. Any users interested in this topic are invited to join the Deep Learning forum.