AI Paper Atlas

Research map / Video / Video segmentation and tracking

Learning Large Motion Estimation from Intermediate Representations with a High-Resolution Optical Flow Dataset Featuring Long-Range Dynamic Motion

Hoonhee Cho, Yuhwan Jeong, Kuk-Jin Yoon

ICCV 2025 · Video › Video segmentation and tracking · approach: Learned optical flow estimation · 0 citations (OpenAlex)

Read the paper

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 the Relay training strategy to RAFTRAFT · RelayFlow-4K testEPE (all regions): 18.54 -> 9.81Improved
Add the Relay training strategy to GMAGMA · RelayFlow-4K testEPE (all regions): 22.08 -> 7.96Improved
Add the Relay strategy to GMFlowNetGMFlowNet · RelayFlow-4K testEPE (all regions): 25.24 -> 13.24Improved
Add the Relay strategy to GMFlowGMFlow · RelayFlow-4K testEPE (all regions): 15.44 -> 8.92Improved
Add Relay to RAFT, long-range flowsRAFT · RelayFlow-4K testEPE s160+: 466.83 -> 230.32Improved
Add Relay to RAFT, unmatched regionsRAFT · RelayFlow-4K testEPE unmatched: 223.82 -> 154.83Improved
Add Relay to GMFlow, short-range flowsGMFlow · RelayFlow-4K testEPE s0-40: 1.99 -> 2.28Hurt
Add Ldis alone to RAFT (ablation)RAFT · RelayFlow-4K testEPE (all regions): 18.54 -> 15.36Improved
Add all of Ldis, incremental time-step learning and LKD (ablation)RAFT · RelayFlow-4K testEPE (all regions): 18.54 -> 9.81Improved
Add Ldis to incremental learning plus LKD (ablation)RAFT · RelayFlow-4K testEPE (all regions): 11.16 -> 9.81Improved
Mask the cost volume with the match map during aggregation (ablation)RAFT-Relay · RelayFlow-4K testEPE (all regions): 12.10 -> 9.81Improved
Train through successive incremental stages 0 to 4 (ablation)RAFT-Relay · RelayFlow-4K testEPE (all regions): 14.04 -> 9.81Improved

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