INGeo: Accelerating Instant Neural Scene Reconstruction with Noisy Geometry Priors

TL;DR

This paper introduces INGeo, a method that leverages noisy geometry priors to significantly accelerate 3D scene reconstruction, aiming for instant training on edge devices like mobile phones and AR/VR headsets.

Contribution

It proposes strategies to mitigate noise in geometry priors, enabling faster training of neural scene reconstruction models on resource-constrained devices.

Findings

Achieves >30 PSNR with half the training iterations on NeRF Synthetic dataset.

Accelerates training speed beyond existing methods on edge devices.

Effectively handles noisy geometry priors to improve reconstruction speed.

Abstract

We present a method that accelerates reconstruction of 3D scenes and objects, aiming to enable instant reconstruction on edge devices such as mobile phones and AR/VR headsets. While recent works have accelerated scene reconstruction training to minute/second-level on high-end GPUs, there is still a large gap to the goal of instant training on edge devices which is yet highly desired in many emerging applications such as immersive AR/VR. To this end, this work aims to further accelerate training by leveraging geometry priors of the target scene. Our method proposes strategies to alleviate the noise of the imperfect geometry priors to accelerate the training speed on top of the highly optimized Instant-NGP. On the NeRF Synthetic dataset, our work uses half of the training iterations to reach an average test PSNR of >30.