AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Optical Flow Training under Limited Label Budget via Active Learning

Shuai Yuan, Xian Sun, Hannah Kim, Shuzhi Yu, Carlo Tomasi

ECCV 2022 · Label-efficient and robust learning › Semi-supervised learning · approach: Pseudo-labeling for specialized tasks · 15 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 20% labels (semi-supervised) to unsupervised ARFlowARFlow · Sintel final valEPE: 2.933 -> 2.598Improved
add 5% labels to unsupervised trainingARFlow · FlyingChairs valEPE: 3.066 -> 2.369Improved
select labels by occ ratio instead of randomSemi-sup ARFlow r=0.1 · Sintel final valEPE: 2.710 -> 2.515Improved
select labels by flow grad norm instead of randomSemi-sup ARFlow r=0.2 · Sintel final valEPE: 2.598 -> 2.299Improved
select labels by photo loss instead of randomSemi-sup ARFlow r=0.2 · Sintel final valEPE: 2.598 -> 2.383Improved
select labels by occ ratio instead of randomSemi-sup ARFlow r=0.05 · KITTI 2012 valFl: 5.363Improved
add unlabeled samples of train split to Stage 2 (semi-sup vs sup only) (ablation)Occ ratio AL r=0.1 · Sintel final valEPE: 2.59 -> 2.52Improved
avoid adding raw Stage-1 unlabeled data to Stage 2 (ablation)Random sampling r=0.1 · Sintel final valEPE: 3.13 -> 2.71Improved
use occ-2x (sample half of top-2x uncertain) instead of occSemi-sup ARFlow r=0.2 · Sintel test finalEPE: 5.09 -> 4.62Improved
use occ ratio instead of randomSemi-sup ARFlow r=0.1 · Sintel test cleanEPE: 3.91 -> 4.28Hurt
use occ ratio instead of randomSemi-sup ARFlow r=0.1 · KITTI-2015 testFl-all: 9.51 -> 8.96Improved
use occ ratio instead of randomSemi-sup ARFlow r=0.2 · KITTI-2015 testFl-all: 8.38 -> 8.30No clear effect

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