DeepStereo: Learning to Predict New Views From the World's Imagery

   Abstract