Few TensoRF: Enhance the Few-shot on Tensorial Radiance Fields

TL;DR

Few TensoRF is a 3D reconstruction framework that combines tensor-based representation with few-shot regularization, achieving fast, high-quality results with sparse input views.

Contribution

It introduces Few TensoRF, enhancing TensorRF with frequency and occlusion masks for improved stability and reconstruction quality in few-shot 3D reconstruction.

Findings

Improves PSNR from 21.45 dB to 23.70 dB on Synthesis NeRF benchmark.

Maintains 10-15 minute training time while enhancing quality.

Achieves 27.37 - 34.00 dB PSNR with only eight images on THuman 2.0.

Abstract

This paper presents Few TensoRF, a 3D reconstruction framework that combines TensorRF's efficient tensor based representation with FreeNeRF's frequency driven few shot regularization. Using TensorRF to significantly accelerate rendering speed and introducing frequency and occlusion masks, the method improves stability and reconstruction quality under sparse input views. Experiments on the Synthesis NeRF benchmark show that Few TensoRF method improves the average PSNR from 21.45 dB (TensorRF) to 23.70 dB, with the fine tuned version reaching 24.52 dB, while maintaining TensorRF's fast \(\approx10-15\) minute training time. Experiments on the THuman 2.0 dataset further demonstrate competitive performance in human body reconstruction, achieving 27.37 - 34.00 dB with only eight input images. These results highlight Few TensoRF as an efficient and data effective solution for real-time 3D reconstruction across diverse scenes.