Unsupervised Learning for Stereo Vision

We consider the problem of learning to estimate depth from stereo image pairs. This can be formulated as unsupervised learning - the training pairs are not labeled with depth. We have formulated an al