SuperGlue: Learning Feature Matching with Graph Neural Networks | VitaVision
PaperCVPR · 2020

SuperGlue: Learning Feature Matching with Graph Neural Networks

Paul-Edouard SarlinDaniel DeToneTomasz MalisiewiczAndrew Rabinovich
arXiv 1911.11763DOIView in graph
1page built on it
4more pages cite it
2narratives
2registry papers cite it

In the Atlas

ModelPrimary source of
SuperGlue

Graph neural network that matches two sets of sparse local features by jointly finding correspondences and rejecting unmatched keypoints in one differentiable forward pass, trained end-to-end with a Sinkhorn optimal-transport assignment over augmented dustbin scores.

Cited as a reference by 4 pages, grouped by domain

Features 3
Feature MatchingLightGlueLoFTR
Representation learning 1
Attention Mechanism

Appears in 2 narratives

NarrativeFoundation Models for VisionNarrativeThe Geometry Pipeline Collapses

Lineage

Papers in the Atlas registry only · one dot per paper
Built upon by · 2
’81’82’83’84’85’86’87’88’89’90’91’92’93’94’95’96’97’98’99’00’01’02’03’04’05’06’07’08’09’10’11’12’13’14’15’16’17’18’19’20’21’22’23LoFTR: Detector-Free Local Feature Matching with Transformers (2021)LightGlue: Local Feature Matching at Light Speed (2023)Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography (1981)Distinctive Image Features from Scale-Invariant Keypoints (2004)ORB: An efficient alternative to SIFT or SURF (2011)SuperPoint: Self-Supervised Interest Point Detection and Description (2018)