DBARF: Deep Bundle-Adjusting Generalizable Neural Radiance Fields
This work addresses a bottleneck in 3D reconstruction for computer vision researchers, enabling more robust and generalizable neural radiance fields across scenes, though it is incremental as it builds on prior bundle adjustment methods.
The paper tackles the problem of jointly optimizing camera poses with Generalizable Neural Radiance Fields (GeNeRFs), which previous methods could not handle, and proposes DBARF, a method that bundle adjusts poses using a cost feature map in a self-supervised manner, achieving effective results on real-world datasets without requiring good initialization.
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}.