Density-aware NeRF Ensembles: Quantifying Predictive Uncertainty in Neural Radiance Fields
This work addresses uncertainty quantification for NeRFs, which is important for applications like view selection and model refinement, but it is incremental as it builds on prior ensembling approaches.
The paper tackles predictive uncertainty in Neural Radiance Fields (NeRFs) by introducing a density-aware ensembling method that accounts for epistemic uncertainty from unobserved scene parts, achieving new state-of-the-art performance on established benchmarks.
We show that ensembling effectively quantifies model uncertainty in Neural Radiance Fields (NeRFs) if a density-aware epistemic uncertainty term is considered. The naive ensembles investigated in prior work simply average rendered RGB images to quantify the model uncertainty caused by conflicting explanations of the observed scene. In contrast, we additionally consider the termination probabilities along individual rays to identify epistemic model uncertainty due to a lack of knowledge about the parts of a scene unobserved during training. We achieve new state-of-the-art performance across established uncertainty quantification benchmarks for NeRFs, outperforming methods that require complex changes to the NeRF architecture and training regime. We furthermore demonstrate that NeRF uncertainty can be utilised for next-best view selection and model refinement.