AI Paper Atlas

Research map / 3D vision / Depth and stereo

Learning End-to-End Scene Flow by Distilling Single Tasks Knowledge

Filippo Aleotti, Matteo Poggi, Fabio Tosi, Stefano Mattoccia

AAAI 2020 · 3D vision › Depth and stereo · approach: Learned optical and scene flow · 37 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Dense connections (ablation)DWARF · FlyingThings3D testFlow EPE: 7.435 -> 6.758Improved
add Dense connections (ablation)DWARF · FlyingThings3D testDisparity EPE: 1.959 -> 1.837Improved
add Dense connections (ablation)DWARF · FlyingThings3D testChange EPE: 2.283 -> 2.092Improved
add 3D correlation to the Dense variant (ablation)DWARF · FlyingThings3D testFlow EPE: 6.758 -> 6.738Improved
add 3D correlation to the Dense variant (ablation)DWARF · FlyingThings3D testDisparity EPE: 1.837 -> 1.827Improved
add 3D correlation to the Dense variant (ablation)DWARF · FlyingThings3D testChange EPE: 2.092 -> 2.149Hurt
add the refinement module to Dense + 3D correlation (ablation)DWARF · FlyingThings3D testFlow EPE: 6.738 -> 6.440Improved
add the refinement module to Dense + 3D correlation (ablation)DWARF · FlyingThings3D testDisparity EPE: 1.827 -> 1.784Improved
add the refinement module to Dense + 3D correlation (ablation)DWARF · FlyingThings3D testChange EPE: 2.149 -> 2.039Improved
use proxy labels only instead of ground-truth fine-tuning (ablation)DWARF · KITTI 2015 validationF1-All: 18.53 -> 20.71Hurt
use proxy labels only instead of ground-truth fine-tuning (ablation)DWARF · KITTI 2015 validationD1-All: 4.58 -> 3.94Improved
fine-tune first on proxy labels, then on ground truth (ablation)DWARF · KITTI 2015 validationF1-All: 18.53 -> 16.75Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy