DWTNeRF: Boosting Few-shot Neural Radiance Fields via Discrete Wavelet Transform

TL;DR

DWTNeRF enhances few-shot neural radiance fields by integrating discrete wavelet loss and multi-head attention, significantly improving convergence speed and rendering quality with sparse views.

Contribution

It introduces a novel discrete wavelet loss and a multi-head attention-based model compatible with Instant-NGP for improved few-shot NeRF performance.

Findings

Outperforms vanilla INGP by 15.07% in PSNR on 3-shot LLFF.

Achieves 24.45% improvement in SSIM.

Reduces overfitting on high frequencies during early training stages.

Abstract

Neural Radiance Fields (NeRF) has achieved superior performance in novel view synthesis and 3D scene representation, but its practical applications are hindered by slow convergence and reliance on dense training views. To this end, we present DWTNeRF, a unified framework based on Instant-NGP's fast-training hash encoding. It is coupled with regularization terms designed for few-shot NeRF, which operates on sparse training views. Our DWTNeRF additionally includes a novel Discrete Wavelet loss that allows explicit prioritization of low frequencies directly in the training objective, reducing few-shot NeRF's overfitting on high frequencies in earlier training stages. We also introduce a model-based approach, based on multi-head attention, that is compatible with INGP, which are sensitive to architectural changes. On the 3-shot LLFF benchmark, DWTNeRF outperforms Vanilla INGP by 15.07% in PSNR, 24.45% in SSIM and 36.30% in LPIPS. Our approach encourages a re-thinking of current few-shot approaches for fast-converging implicit representations like INGP or 3DGS.