AI Paper Atlas

Research map / 3D vision / Depth and stereo

ProbFlow: Joint Optical Flow and Uncertainty Estimation

Anne S. Wannenwetsch, Margret Keuper, Stefan Roth

ICCV 2017 · 3D vision › Depth and stereo · approach: Learned optical and scene flow · 56 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
replace best post-hoc measure (Learned) with ProbFlow marginal entropyuncertainty estimate · Middlebury trainingAUC: 0.496 -> 0.466Improved
replace Learned uncertainty with ProbFlow marginal entropyuncertainty estimate · Middlebury trainingSpearman CC: 0.324 -> 0.374Improved
replace Energy/Learned uncertainty with ProbFlowFields entropyuncertainty estimate · Sintel validationAUC: 0.474 -> 0.398Improved
replace Learned uncertainty with ProbFlowFields entropyuncertainty estimate · Sintel validationSpearman CC: 0.451 -> 0.563Improved
replace MAP FlowFields* with probabilistic inference (ProbFlowFields)FlowFields energy · Sintel validationAEPE: 3.161 -> 3.127Improved
use probabilistic ClassicA instead of ClassicAClassicA energy · Middlebury trainingAEPE: 0.295 -> 0.296No clear effect
apply bilateral solver guided by uncertainties on ProbFlowFieldsProbFlowFields · Sintel validationAEPE: 3.127 -> 3.052Improved
add uncertainty-based track termination (3) to FB check (1) (ablation)trajectory generation · FBMS-59 trainingF-measure: 78.03% -> 80.92%Improved
use ProbFlowFields instead of FlowFields in trajectory generationmotion segmentation · FBMS-59 trainingpoint density: 0.83% -> 1.18%Improved
jointly estimate uncertainty (ProbFlowFields) instead of FlowFields plus learned post-hoc measureruntime · Sintel validationruntime (s): 123.8s -> 38.1sImproved
add uncertainty estimation to FlowFieldsruntime · Sintel validationruntime (s): 19.9s -> 38.1sHurt
use ProbFlowFields dense segmentation instead of FlowFields densemotion segmentation · FBMS-59 test (dense)F-measure: 70.83% -> 75.05%Improved

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