Digging Into Self-Supervised Monocular Depth Estimation: Difference between revisions

Created page with "Digging Into Self-Supervised Monocular Depth Estimation (ICCV 2019) Monodepth2 Authors: Clement Godard, Oisin Mac Aodha, Michael Firman, Gabriel Brostow Affiliations: UCL,..."
 
No edit summary
 
(One intermediate revision by the same user not shown)
Line 5: Line 5:
Affiliations: UCL, Caltech, Niantic
Affiliations: UCL, Caltech, Niantic


* [https://arxiv.org/abs/1806.01260 Arxiv mirror]
* [https://arxiv.org/abs/1806.01260 Arxiv mirror] [https://openaccess.thecvf.com/content_ICCV_2019/html/Godard_Digging_Into_Self-Supervised_Monocular_Depth_Estimation_ICCV_2019_paper.html CVF Mirror]
* [https://github.com/nianticlabs/monodepth2 Github]
* [https://github.com/nianticlabs/monodepth2 Github]


Line 20: Line 20:


===Per-Pixel Minimum Reprojection Loss===
===Per-Pixel Minimum Reprojection Loss===
Basically you have two images in a sequence: frame1, frame2, frame3. 
Each gives you a loss:
<pre>
loss1 = abs(frame2 - warp(frame1))
loss2 = abs(frame2 - warp(frame3))
# Take the minimum over all pixels
loss = mean(min(loss1, loss2))
</pre>
===Auto-Masking Stationary Pixels===
===Auto-Masking Stationary Pixels===
===Multi-scale Estimation===
===Multi-scale Estimation===