Bayes' Rays: Uncertainty Quantification for Neural Radiance Fields

TL;DR

BayesRays provides a novel, efficient post-hoc method for quantifying uncertainty in pre-trained Neural Radiance Fields, improving reliability in view synthesis and depth estimation tasks.

Contribution

It introduces a Bayesian Laplace approximation-based framework that evaluates uncertainty in NeRFs without retraining, enhancing practical applicability.

Findings

Outperforms existing heuristic uncertainty methods

Accurately quantifies uncertainty in view synthesis

Applicable to any pre-trained NeRF model

Abstract

Neural Radiance Fields (NeRFs) have shown promise in applications like view synthesis and depth estimation, but learning from multiview images faces inherent uncertainties. Current methods to quantify them are either heuristic or computationally demanding. We introduce BayesRays, a post-hoc framework to evaluate uncertainty in any pre-trained NeRF without modifying the training process. Our method establishes a volumetric uncertainty field using spatial perturbations and a Bayesian Laplace approximation. We derive our algorithm statistically and show its superior performance in key metrics and applications. Additional results available at: https://bayesrays.github.io.