AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

S2DNet: Learning Image Features for Accurate Sparse-to-Dense Matching

Hugo Germain, Guillaume Bourmaud, Vincent Lepetit

ECCV 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 73 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
use sparse-to-dense instead of sparse-to-sparse matching (ablation)S2DNet + SuperPoint · HPatchesMMA@1px: 0.373 -> 0.563Improved
use sparse-to-dense instead of sparse-to-sparse matching (ablation)S2DNet + D2-Net detector · HPatchesMMA@1px: 0.118 -> 0.467Improved
add confidence thresholding tau = 0.20 (ablation)S2DNet + SuperPoint · HPatchesMMA@1px: 0.469 -> 0.563Improved
add confidence thresholding tau = 0.20 (ablation)S2DNet + Harris · HPatchesMMA@1px: 0.441 -> 0.511Improved
use SuperPoint instead of Harris detector for source keypoints (ablation)S2DNet · HPatchesMMA@1px: 0.511 -> 0.563Improved
use sparse-to-dense with D2-Net and R2D2 features under illumination change (ablation)D2-Net and R2D2 · HPatches illuminationMMAImproved
use sparse-to-dense with D2-Net and R2D2 features under viewpoint change (ablation)D2-Net and R2D2 · HPatches viewpointMMAHurt
run weakly supervised S2DHM in sparse-to-sparse setting (ablation)S2DHM · HPatchesMMAImproved
replace weakly supervised features with supervised classification-trained S2DNet featuresS2DHM · Aachen Nightrecall 0.25m/2deg: 30.6 -> 46.9Improved
use S2DNet instead of D2-Net on day localizationD2-Net · RobotCar Seasons Day-Allrecall 0.25m/2deg: 54.5 -> 53.9No clear effect
use S2DNet instead of D2-Net on nighttime RobotCar finest thresholdD2-Net · RobotCar Night-Allrecall 0.25m/2deg: 20.4 -> 14.5Hurt
use S2DNet instead of D2-Net on InLoc direct PE medium and coarse thresholdsD2-Net · InLoc Direct PErecall 5m/10deg: 48.6 -> 48.5No 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