I2VGuard: Safeguarding Images against Misuse in Diffusion-based Image-to-Video Models

Dongnan Gui; Xun Guo; Wengang Zhou; Yan Lu

doi:10.1109/cvpr52734.2025.01175

CVPR2025顶会

I2VGuard: Safeguarding Images against Misuse in Diffusion-based Image-to-Video Models

Dongnan Gui, Xun Guo, Wengang Zhou, Yan Lu

DOI 出版方

2025年份

3顶会引用

摘要

Original Image Generated video from original image. Guarded Image Generated video from guarded image. Original Image Generated video from original image. Guarded Image Generated video from guarded image. Figure 1 . Results of our I2VGuard. We present original images, guarded images, and their corresponding SVD-generated videos. All results are generated with the same seed. Our method effectively safeguards images from animation in image-to-video generation.