CROSSFIRE: Camera Relocalization On Self-Supervised Features from an Implicit Representation
This addresses the problem of robust camera localization for navigation in dynamic environments, offering a novel integration with neural renderers, though it builds incrementally on existing implicit representation techniques.
The paper tackles camera relocalization in real-world scenes by using Neural Radiance Fields as an implicit map, proposing a method that computes precise device positions in real-time from a single RGB camera. The result is improved accuracy over competitors, with the ability to handle dynamic outdoor environments and changing lighting conditions.
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.