AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

ScopeFlow: Dynamic Scene Scoping for Optical Flow

Aviram Bar-Haim, Lior Wolf

CVPR 2020 · Architectures and efficiency › RNNs and normalization · approach: Dataset condensation and flow estimation · 60 citations (OpenAlex)

Check 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
Replace fixed partial crops with randomized crop sizes in [0.95,1] (ablation)IRR-PWC · KITTI validationMEPE: 1.651 -> 1.402Improved
Use random-range crops [0.95,1] instead of fixed full crops (ablation)IRR-PWC · KITTI validationMEPE: 1.594 -> 1.402Improved
Re-finetune from the [0.9,1] crop-range checkpoint rather than the [0.95,1] checkpoint (ablation)IRR-PWC on KITTI · KITTI validationMEPE: 1.393 -> 1.377Improved
Remove weight decay and random noise during early FChairsOcc pretraining (ablation)IRR-PWC · Sintel testMEPE: 3.138 -> 3.264Hurt
Reduce zoom-in during early FChairsOcc pretraining (ablation)IRR-PWC · Sintel testMEPE: 3.138 -> 3.321Hurt
Increase scope with maximal crops during FThings training, starting from C1 rather than C* (ablation)IRR-PWC · Sintel testMEPE: 2.613 -> 2.544Improved
Remove weight decay and random noise during additional FThings training (ablation)IRR-PWC · Sintel testMEPE: 2.544 -> 2.545No clear effect
Remove weight decay and random noise during Sintel finetuning (ablation)IRR-PWC · Sintel validationMEPE: 2.119 -> 2.108Improved
Remove weight decay and random noise during Sintel finetuning (ablation)IRR-PWC · Sintel validationocclusion F1: 0.700 -> 0.703Improved
Train IRR-PWC with the ScopeFlow protocol rather than the baseline protocolIRR-PWC · MPI Sintel final passall EPE: 4.579 -> 4.098Improved
Train IRR-PWC with the ScopeFlow protocol rather than the baseline protocolIRR-PWC · KITTI 2012Avg-All: 1.6 -> 1.3Improved
Train IRR-PWC with the ScopeFlow protocol rather than the baseline protocolIRR-PWC · KITTI 2015Fl-all: 7.65 -> 6.82Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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