Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PRIME: Protect Your Videos From Malicious Editing

Guanlin Li, Shuai Yang, Jie Zhang, Tianwei Zhang

TL;DR

PRIME addresses the risk of malicious video editing enabled by latent diffusion models by providing a black-box protection method that perturbs every video frame with zero-shot transferability. It introduces fast convergence searching, early stopping, and anti dynamic compression to achieve per-frame perturbations that survive codec compression and generalize across editing pipelines. Quantitative and human evaluations on the VIOLENT dataset show PRIME reduces the quality of edited videos and is far more efficient (8.3% GPU hours) than the prior Photoguard, while increasing bitrate modestly. The work offers practical defenses for protecting portrait rights in user-generated videos and highlights ethical and policy considerations around generative video misuse.

Abstract

With the development of generative models, the quality of generated content keeps increasing. Recently, open-source models have made it surprisingly easy to manipulate and edit photos and videos, with just a few simple prompts. While these cutting-edge technologies have gained popularity, they have also given rise to concerns regarding the privacy and portrait rights of individuals. Malicious users can exploit these tools for deceptive or illegal purposes. Although some previous works focus on protecting photos against generative models, we find there are still gaps between protecting videos and images in the aspects of efficiency and effectiveness. Therefore, we introduce our protection method, PRIME, to significantly reduce the time cost and improve the protection performance. Moreover, to evaluate our proposed protection method, we consider both objective metrics and human subjective metrics. Our evaluation results indicate that PRIME only costs 8.3% GPU hours of the cost of the previous state-of-the-art method and achieves better protection results on both human evaluation and objective metrics. Code can be found in https://github.com/GuanlinLee/prime.

PRIME: Protect Your Videos From Malicious Editing

TL;DR

PRIME addresses the risk of malicious video editing enabled by latent diffusion models by providing a black-box protection method that perturbs every video frame with zero-shot transferability. It introduces fast convergence searching, early stopping, and anti dynamic compression to achieve per-frame perturbations that survive codec compression and generalize across editing pipelines. Quantitative and human evaluations on the VIOLENT dataset show PRIME reduces the quality of edited videos and is far more efficient (8.3% GPU hours) than the prior Photoguard, while increasing bitrate modestly. The work offers practical defenses for protecting portrait rights in user-generated videos and highlights ethical and policy considerations around generative video misuse.

Abstract

With the development of generative models, the quality of generated content keeps increasing. Recently, open-source models have made it surprisingly easy to manipulate and edit photos and videos, with just a few simple prompts. While these cutting-edge technologies have gained popularity, they have also given rise to concerns regarding the privacy and portrait rights of individuals. Malicious users can exploit these tools for deceptive or illegal purposes. Although some previous works focus on protecting photos against generative models, we find there are still gaps between protecting videos and images in the aspects of efficiency and effectiveness. Therefore, we introduce our protection method, PRIME, to significantly reduce the time cost and improve the protection performance. Moreover, to evaluate our proposed protection method, we consider both objective metrics and human subjective metrics. Our evaluation results indicate that PRIME only costs 8.3% GPU hours of the cost of the previous state-of-the-art method and achieves better protection results on both human evaluation and objective metrics. Code can be found in https://github.com/GuanlinLee/prime.
Paper Structure (18 sections, 5 equations, 1 figure, 4 tables, 1 algorithm)

This paper contains 18 sections, 5 equations, 1 figure, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Video Editing with LDMs
  4. Image Protection Against LDMs-based Misuse
  5. Preliminary
  6. Protect Videos from Malicious Editing
  7. Motivation
  8. Threat Model
  9. PRIME
  10. Experiments
  11. Data Collection
  12. Experiment Settings
  13. Metrics
  14. Protection Evaluation
  15. Conclusion
  16. ...and 3 more sections

Figures (1)

  • Figure 1: Human evaluation on six subjective metrics for two malicious editing tasks. We show the mean score for each metric and put the standard deviation in brackets. The results are calculated on the edited videos.