Invisible Backdoor Triggers in Image Editing Model via Deep Watermarking
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909639282524160 |
|---|---|
| author | Chen, Yu-Feng Huang, Tzuhsuan Chiu, Pin-Yen Chen, Jun-Cheng |
| author_facet | Chen, Yu-Feng Huang, Tzuhsuan Chiu, Pin-Yen Chen, Jun-Cheng |
| contents | Diffusion models have achieved remarkable progress in both image generation and editing. However, recent studies have revealed their vulnerability to backdoor attacks, in which specific patterns embedded in the input can manipulate the model's behavior. Most existing research in this area has proposed attack frameworks focused on the image generation pipeline, leaving backdoor attacks in image editing relatively unexplored. Among the few studies targeting image editing, most utilize visible triggers, which are impractical because they introduce noticeable alterations to the input image before editing. In this paper, we propose a novel attack framework that embeds invisible triggers into the image editing process via poisoned training data. We leverage off-the-shelf deep watermarking models to encode imperceptible watermarks as backdoor triggers. Our goal is to make the model produce the predefined backdoor target when it receives watermarked inputs, while editing clean images normally according to the given prompt. With extensive experiments across different watermarking models, the proposed method achieves promising attack success rates. In addition, the analysis results of the watermark characteristics in term of backdoor attack further support the effectiveness of our approach. The code is available at:https://github.com/aiiu-lab/BackdoorImageEditing |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_04879 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Invisible Backdoor Triggers in Image Editing Model via Deep Watermarking Chen, Yu-Feng Huang, Tzuhsuan Chiu, Pin-Yen Chen, Jun-Cheng Computer Vision and Pattern Recognition Diffusion models have achieved remarkable progress in both image generation and editing. However, recent studies have revealed their vulnerability to backdoor attacks, in which specific patterns embedded in the input can manipulate the model's behavior. Most existing research in this area has proposed attack frameworks focused on the image generation pipeline, leaving backdoor attacks in image editing relatively unexplored. Among the few studies targeting image editing, most utilize visible triggers, which are impractical because they introduce noticeable alterations to the input image before editing. In this paper, we propose a novel attack framework that embeds invisible triggers into the image editing process via poisoned training data. We leverage off-the-shelf deep watermarking models to encode imperceptible watermarks as backdoor triggers. Our goal is to make the model produce the predefined backdoor target when it receives watermarked inputs, while editing clean images normally according to the given prompt. With extensive experiments across different watermarking models, the proposed method achieves promising attack success rates. In addition, the analysis results of the watermark characteristics in term of backdoor attack further support the effectiveness of our approach. The code is available at:https://github.com/aiiu-lab/BackdoorImageEditing |
| title | Invisible Backdoor Triggers in Image Editing Model via Deep Watermarking |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.04879 |