DeepStereo: Google quartet has method for new-view synthesis
Thursday, July 9, 2015 - 04:30
in Mathematics & Economics
Four researchers from Google have been turning to deep networks—not for classification purposes in computer vision but this time for application in graphics problems. Their work has shown interesting results, as evidenced in their paper, now on the arXiv server, titled "DeepStereo: Learning to Predict New Views from the World's Imagery."