SuperGlue: Learning Feature Matching with Graph Neural Networks

TL;DR

SuperGlue is a neural network that improves feature matching in images by jointly estimating correspondences and rejecting non-matches using graph neural networks and optimal transport, achieving state-of-the-art results in pose estimation.

Contribution

It introduces a novel graph neural network-based approach for feature matching that jointly reasons about scene geometry and feature correspondences in an end-to-end trainable manner.

Findings

Outperforms previous learned feature matching methods.

Achieves state-of-the-art pose estimation accuracy.

Operates in real-time on modern GPUs.

Abstract

This paper introduces SuperGlue, a neural network that matches two sets of local features by jointly finding correspondences and rejecting non-matchable points. Assignments are estimated by solving a differentiable optimal transport problem, whose costs are predicted by a graph neural network. We introduce a flexible context aggregation mechanism based on attention, enabling SuperGlue to reason about the underlying 3D scene and feature assignments jointly. Compared to traditional, hand-designed heuristics, our technique learns priors over geometric transformations and regularities of the 3D world through end-to-end training from image pairs. SuperGlue outperforms other learned approaches and achieves state-of-the-art results on the task of pose estimation in challenging real-world indoor and outdoor environments. The proposed method performs matching in real-time on a modern GPU and can be readily integrated into modern SfM or SLAM systems. The code and trained weights are publicly available at https://github.com/magicleap/SuperGluePretrainedNetwork.