Like iWildCam 2018, this years' competition will be based on data from camera traps: cameras used by biologists to study animals in the wild. iWildCam 2019 will challenge competitors to face the real-world open set problem that occurs when a camera trap is placed in a new geographic region, one where the set of possible species does not perfectly match species seen in previous camera trap data.
Checkout the iWildCam Competition Github repo for the specifics of the dataset and download links. The competition itself is being hosted on Kaggle.
Competition Begins 18 March 2019
Submission Deadline 7 June 2019
Sara Beery (Caltech), Dan Morris (Microsoft)
Microsoft AI for Earth, National Park Service, United States Geological Survey, and Idaho Department of Fish and Game