Evidential Neural Radiance Fields
This work addresses uncertainty quantification for NeRFs, a problem for safety-critical 3D scene modeling, and is incremental as it builds on existing NeRF methods.
The paper tackles the lack of uncertainty estimation in neural radiance fields (NeRFs), which limits their use in safety-critical applications, by introducing Evidential Neural Radiance Fields, a method that quantifies both aleatoric and epistemic uncertainty in a single forward pass without compromising rendering quality or adding computational overhead, achieving state-of-the-art results on three benchmarks.
Understanding sources of uncertainty is fundamental to trustworthy three-dimensional scene modeling. While recent advances in neural radiance fields (NeRFs) achieve impressive accuracy in scene reconstruction and novel view synthesis, the lack of uncertainty estimation significantly limits their deployment in safety-critical settings. Existing uncertainty quantification methods for NeRFs fail to capture both aleatoric and epistemic uncertainty. Among those that do quantify one or the other, many of them either compromise rendering quality or incur significant computational overhead to obtain uncertainty estimates. To address these issues, we introduce Evidential Neural Radiance Fields, a probabilistic approach that seamlessly integrates with the NeRF rendering process and enables direct quantification of both aleatoric and epistemic uncertainty from a single forward pass. We compare multiple uncertainty quantification methods on three standardized benchmarks, where our approach demonstrates state-of-the-art scene reconstruction fidelity and uncertainty estimation quality.