Abstract: Self-supervised monocular depth estimation leverages photometric consistency across views as supervisory signals, yet inherently suffers from compromised geometric fidelity at object ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results