Even today’s best #deeplearning vision systems can fail when pose/lighting/background vary. Our work on ImageNetX is one of the first large scale efforts to pinpoint mistake types of in AI computer vision systems. Explore the dataset

→ View original post on X — @aiatmeta