DBARF: Deep Bundle-Adjusting Generalizable Neural Radiance Fields

TL;DR

DBARF introduces a novel method for jointly optimizing camera poses with generalizable neural radiance fields (GeNeRFs), enabling scene-agnostic, self-supervised training without requiring accurate initial poses.

Contribution

We propose DBARF, a new approach that bundle adjusts camera poses using a cost feature map, allowing joint training with GeNeRFs across scenes without initial pose estimates.

Findings

Effective on real-world datasets

Generalizes across different scenes

Does not require accurate initial camera poses

Abstract

Recent works such as BARF and GARF can bundle adjust camera poses with neural radiance fields (NeRF) which is based on coordinate-MLPs. Despite the impressive results, these methods cannot be applied to Generalizable NeRFs (GeNeRFs) which require image feature extractions that are often based on more complicated 3D CNN or transformer architectures. In this work, we first analyze the difficulties of jointly optimizing camera poses with GeNeRFs, and then further propose our DBARF to tackle these issues. Our DBARF which bundle adjusts camera poses by taking a cost feature map as an implicit cost function can be jointly trained with GeNeRFs in a self-supervised manner. Unlike BARF and its follow-up works, which can only be applied to per-scene optimized NeRFs and need accurate initial camera poses with the exception of forward-facing scenes, our method can generalize across scenes and does not require any good initialization. Experiments show the effectiveness and generalization ability of our DBARF when evaluated on real-world datasets. Our code is available at \url{https://aibluefisher.github.io/dbarf}.