Online Invariance Selection for Local Feature Descriptors

TL;DR

This paper introduces LISRD, a method that learns multiple local descriptors with varying invariance levels and dynamically selects the most suitable one at runtime, improving matching performance under challenging conditions.

Contribution

It proposes a novel framework for disentangling invariance in local descriptors and online selection based on regional variations, enhancing robustness and discriminative power.

Findings

Outperforms state-of-the-art descriptors on challenging datasets.

Boosts matching accuracy under illumination and viewpoint changes.

Enables adaptive invariance selection for local features.

Abstract

To be invariant, or not to be invariant: that is the question formulated in this work about local descriptors. A limitation of current feature descriptors is the trade-off between generalization and discriminative power: more invariance means less informative descriptors. We propose to overcome this limitation with a disentanglement of invariance in local descriptors and with an online selection of the most appropriate invariance given the context. Our framework consists in a joint learning of multiple local descriptors with different levels of invariance and of meta descriptors encoding the regional variations of an image. The similarity of these meta descriptors across images is used to select the right invariance when matching the local descriptors. Our approach, named Local Invariance Selection at Runtime for Descriptors (LISRD), enables descriptors to adapt to adverse changes in images, while remaining discriminative when invariance is not required. We demonstrate that our method can boost the performance of current descriptors and outperforms state-of-the-art descriptors in several matching tasks, when evaluated on challenging datasets with day-night illumination as well as viewpoint changes.