CROSSFIRE: Camera Relocalization On Self-Supervised Features from an Implicit Representation

TL;DR

This paper introduces a real-time camera relocalization method using self-supervised features from Neural Radiance Fields, achieving high accuracy in dynamic outdoor environments without relying on pose regression.

Contribution

It presents a novel relocalization algorithm leveraging implicit scene representations and self-supervised dense features, avoiding pose regression and photometric alignment.

Findings

More accurate than existing methods

Operates effectively in dynamic outdoor environments

Can be integrated with volumetric neural renderers

Abstract

Beyond novel view synthesis, Neural Radiance Fields are useful for applications that interact with the real world. In this paper, we use them as an implicit map of a given scene and propose a camera relocalization algorithm tailored for this representation. The proposed method enables to compute in real-time the precise position of a device using a single RGB camera, during its navigation. In contrast with previous work, we do not rely on pose regression or photometric alignment but rather use dense local features obtained through volumetric rendering which are specialized on the scene with a self-supervised objective. As a result, our algorithm is more accurate than competitors, able to operate in dynamic outdoor environments with changing lightning conditions and can be readily integrated in any volumetric neural renderer.