ContextDesc: Local Descriptor Augmentation with Cross-Modality Context

TL;DR

This paper introduces ContextDesc, a novel framework that enhances local feature descriptors with cross-modality contextual information, improving geometric matching performance across diverse scenes.

Contribution

It proposes a unified learning framework that combines visual and geometric context for local descriptors and introduces an N-pair loss for better convergence.

Findings

Significant improvement on large-scale benchmarks

Lightweight augmentation scheme

Enhanced generalization in geometric matching

Abstract

Most existing studies on learning local features focus on the patch-based descriptions of individual keypoints, whereas neglecting the spatial relations established from their keypoint locations. In this paper, we go beyond the local detail representation by introducing context awareness to augment off-the-shelf local feature descriptors. Specifically, we propose a unified learning framework that leverages and aggregates the cross-modality contextual information, including (i) visual context from high-level image representation, and (ii) geometric context from 2D keypoint distribution. Moreover, we propose an effective N-pair loss that eschews the empirical hyper-parameter search and improves the convergence. The proposed augmentation scheme is lightweight compared with the raw local feature description, meanwhile improves remarkably on several large-scale benchmarks with diversified scenes, which demonstrates both strong practicality and generalization ability in geometric matching applications.