Unsupervised monocular depth estimation by optimizing with knowledge from Flow-Net with FG-Depth

Unsupervised monocular depth estimation by optimizing with knowledge from Flow-Net with FG-Depth

FG-Depth: Flow-Guided Unsupervised Monocular Depth Estimation
arXiv paper abstract https://arxiv.org/abs/2301.08414
arXiv PDF paper https://arxiv.org/pdf/2301.08414.pdf

The great potential of unsupervised monocular depth estimation has been demonstrated by many works due to low annotation cost and impressive accuracy comparable to supervised methods.

… However, previous methods prove that this image reconstruction optimization is prone to get trapped in local minima.

… core idea is to guide the optimization with prior knowledge from pretrained Flow-Net.

… show that the bottleneck of unsupervised monocular depth estimation can be broken with … simple but effective framework named FG-Depth.

… propose (i) a flow distillation loss to replace the typical photometric loss that limits the capacity of the model and (ii) a prior flow based mask to remove invalid pixels that bring the noise in training loss.

… approach achieves state-of-the-art results on both KITTI and NYU-Depth-v2 datasets.

Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website

LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b

Photo by Matt Seymour on Unsplash