AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Fast Optical Flow Using Dense Inverse Search

Till Kroeger, Radu Timofte, Dengxin Dai, Luc Van Gool

ECCV 2016 · Architectures and efficiency › RNNs and normalization · approach: Dataset condensation and flow estimation · 393 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 densification (patch averaging weighted by photometric error) between scales (ablation)DIS inverse search · Sintel training sparse correspondencesEPE all (px): 7.76 -> 4.16Improved
add densification (ablation)DIS inverse search · Sintel training, displacement 40+ pxEPE (px): 37.94 -> 23.09Improved
add densification (ablation)DIS inverse search · Sintel training, displacement 0-10 pxEPE (px): 2.16 -> 0.84Improved
use inverse search with densification instead of exhaustive NCC nearest-neighborDIS · Sintel training sparse correspondencesEPE all (px): 32.06 -> 4.16Improved
use DeepMatching instead of DISsparse matching · Sintel training sparse correspondencesEPE all (px): 4.16 -> 3.60Improved
add variational refinement (ablation)DIS · Sintel trainingaverage EPEImproved
increase finest-scale resolution, iterations or patch overlap (ablation)DIS · Sintel trainingaverage EPEImproved
vary patch size theta_ps (ablation)DIS · Sintel trainingaverage EPEMixed
add OpenMP parallelization on 4 cores (ablation)DIS · Sintel run-timespeed-up (x): 1.29 -> 3.70Improved
run DIS at higher frame rate through all intermediate frames instead of DeepFlow on large displacementsDIS operating point 3 · Sintel with synthesized low frame-rateEPEImproved
remove variational refinement (operating point 1 vs 2) (ablation)DIS · Sintelrun-time: 301 Hz -> 606 HzImproved

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