Code for unsupervised learning of object landmarks as proposed in "Unsupervised Learning of Object Landmarks through Conditional Image Generation", Tomas Jakab*, Ankush Gupta*, Hakan Bilen, Andrea Vedaldi, Advances in Neural Information Processing Systems (NeurIPS) 2018.
Can you provide the loss statistics for training? Logs would be helpful.