AegisRF: Adversarial Perturbations Guided with Sensitivity for Protecting Intellectual Property of Neural Radiance Fields


Woo Jae Kim (Korea Advanced Institute of Science and Technology (KAIST)), Kyu Beom Han (Korea Advanced Institute of Science and Technology (KAIST)), Yoonki Cho (Korea Advanced Institute of Science and Technology (KAIST)), Youngju Na (Korea Advanced Institute of Science and Technology (KAIST)), Junsik Jung (Korea Advanced Institute of Science and Technology (KAIST)), Sooel Son (Korea Advanced Institute of Science and Technology (KAIST)), Sung-Eui Yoon (Korea Advanced Institute of Science and Technology (KAIST))
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

As Neural Radiance Fields (NeRFs) have emerged as a powerful tool for 3D scene representation and novel view synthesis, protecting their intellectual property (IP) from unauthorized use is becoming increasingly crucial. In this work, we aim to protect the IP of NeRFs by injecting adversarial perturbations that disrupt their unauthorized applications. However, perturbing the 3D geometry of NeRFs can easily deform the underlying scene structure and thus substantially degrade the rendering quality, which has led existing attempts to avoid geometric perturbations or restrict them to explicit spaces like meshes. To overcome this limitation, we introduce a learnable sensitivity to quantify the spatially varying impact of geometric perturbations on rendering quality. Building upon this, we propose AegisRF, a novel framework that consists of a Perturbation Field, which injects adversarial perturbations into the pre-rendering outputs (color and volume density) of NeRF models to fool an unauthorized downstream target model, and a Sensitivity Field, which learns the sensitivity to adaptively constrain geometric perturbations, preserving rendering quality while disrupting unauthorized use. Our experimental evaluations demonstrate the generalized applicability of AegisRF across diverse downstream tasks and modalities, including multi-view image classification and voxel-based 3D localization, while maintaining high visual fidelity.

Citation

@inproceedings{Kim_2025_BMVC,
author    = {Woo Jae Kim and Kyu Beom Han and Yoonki Cho and Youngju Na and Junsik Jung and Sooel Son and Sung-Eui Yoon},
title     = {AegisRF: Adversarial Perturbations Guided with Sensitivity for Protecting Intellectual Property of Neural Radiance Fields},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_283/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection