PH-Dropout: Practical Epistemic Uncertainty Quantification for View Synthesis

TL;DR

This paper introduces PH-Dropout, a real-time method for epistemic uncertainty quantification in view synthesis models like NeRF and GS, addressing a critical gap for robustness and scalability.

Contribution

We propose PH-Dropout, the first efficient post hoc uncertainty estimation method for pre-trained NeRF and GS models, enhancing robustness and scalability in view synthesis.

Findings

PH-Dropout provides accurate uncertainty estimates in real-time.

It outperforms existing methods in computational efficiency.

Validated through extensive evaluations on real-world data.

Abstract

View synthesis using Neural Radiance Fields (NeRF) and Gaussian Splatting (GS) has demonstrated impressive fidelity in rendering real-world scenarios. However, practical methods for accurate and efficient epistemic Uncertainty Quantification (UQ) in view synthesis are lacking. Existing approaches for NeRF either introduce significant computational overhead (e.g., ``10x increase in training time" or ``10x repeated training") or are limited to specific uncertainty conditions or models. Notably, GS models lack any systematic approach for comprehensive epistemic UQ. This capability is crucial for improving the robustness and scalability of neural view synthesis, enabling active model updates, error estimation, and scalable ensemble modeling based on uncertainty. In this paper, we revisit NeRF and GS-based methods from a function approximation perspective, identifying key differences and connections in 3D representation learning. Building on these insights, we introduce PH-Dropout (Post hoc Dropout), the first real-time and accurate method for epistemic uncertainty estimation that operates directly on pre-trained NeRF and GS models. Extensive evaluations validate our theoretical findings and demonstrate the effectiveness of PH-Dropout.