How GPUs and Deep Learning Are Helping Protect Endangered Species

Finding a cheetah across a dozen square miles of African bush can be difficult. But that same animal might lay down 200,000 footprints in a day. So finding its tracks is a lot easier. And it turns out that, beyond being useful for locating animals, tracks can tell you a whole lot more. This insight has led to a potentially revolutionary approach to wildlife conservation that’s about to be jump-started by deep learning and NVIDIA GPUs. Zoe Jewell and Sky Alibhai are co-founders of WildTrack, a nonprofit devoted to monitoring endangered species. In time for World Environment Day, they’ve launched a program called ConservationFIT, where FIT stands for “footprint identification technology.” The program seeks to crowdsource photos of animal footprints, and then use those images to build algorithms that can identify the species, individual, sex and age-class of the animal who made them.