arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2406.04101cs.CV
收藏

我们能将基于 Instant-NGP 的 NeRF 压缩到什么程度?

How Far Can We Compress Instant-NGP-Based NeRF?

Shanghai Jiao Tong University(上海交通大学);Monash University(莫纳什大学)

Yihang Chen, Qianyi Wu, Mehrtash Harandi, Jianfei Cai

AI总结:

本文提出基于上下文的 NeRF 压缩框架 CNC,通过层级与维度上下文建模及哈希碰撞和占用网格先验,在保持保真度的同时大幅压缩 Instant-NGP 类 NeRF 的存储体积。

摘要:

近年来,神经辐射场(NeRF)在表示三维场景方面展现出卓越的能力。为加速渲染过程,可学习的显式表示被引入并与隐式 NeRF 表示相结合,但这导致存储空间需求大幅增加。本文提出基于上下文的 NeRF 压缩(Context-based NeRF Compression,CNC)框架,利用高效的上下文模型提供存储友好的 NeRF 表示。具体而言,我们挖掘层级维度和通道维度上的上下文依赖关系,以实现信息熵降低的概率预测。此外,我们利用哈希碰撞和占用网格作为强先验知识,以改进上下文建模。据我们所知,我们是首个为 NeRF 压缩构建并利用上下文模型的工作。在 Synthetic-NeRF 和 Tanks and Temples 数据集上,与基线 Instant-NGP 相比,我们分别实现了 100 倍和 70 倍的体积缩减,同时保真度有所提升。此外,与当前最优的 NeRF 压缩方法 BiRF 相比,我们分别实现了 86.7% 和 82.3% 的存储空间缩减。代码已开源:https://github.com/YihangChen-ee/CNC。

英文摘要:

In recent years, Neural Radiance Field (NeRF) has demonstrated remarkable capabilities in representing 3D scenes. To expedite the rendering process, learnable explicit representations have been introduced for combination with implicit NeRF representation, which however results in a large storage space requirement. In this paper, we introduce the Context-based NeRF Compression (CNC) framework, which leverages highly efficient context models to provide a storage-friendly NeRF representation. Specifically, we excavate both level-wise and dimension-wise context dependencies to enable probability prediction for information entropy reduction. Additionally, we exploit hash collision and occupancy grids as strong prior knowledge for better context modeling. To the best of our knowledge, we are the first to construct and exploit context models for NeRF compression. We achieve a size reduction of 100$\times$ and 70$\times$ with improved fidelity against the baseline Instant-NGP on Synthesic-NeRF and Tanks and Temples datasets, respectively. Additionally, we attain 86.7\% and 82.3\% storage size reduction against the SOTA NeRF compression method BiRF. Our code is available here: https://github.com/YihangChen-ee/CNC.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑