Creating Blank Canvas Against AI-enabled Image Forgery

2025/11/27 by Song Qi, Song, Qi, Ziyuan Luo +3
Computer Science · #Advanced Image Processing Techniques #Computer Vision and Pattern Recognition (cs.CV) #Digital Media Forensic Detection #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis

paper · pdf · doi:10.48550/arxiv.2511.22237

openalex publication_date 2025/11/27 · openalex created_date 2025/12/03 · openalex updated_date 2026/07/28

Abstract

AIGC-based image editing technology has greatly simplified the realistic-level image modification, causing serious potential risks of image forgery. This paper introduces a new approach to tampering detection using the Segment Anything Model (SAM). Instead of training SAM to identify tampered areas, we propose a novel strategy. The entire image is transformed into a blank canvas from the perspective of neural models. Any modifications to this blank canvas would be noticeable to the models. To achieve this idea, we introduce adversarial perturbations to prevent SAM from ``seeing anything'', allowing it to identify forged regions when the image is tampered with. Due to SAM's powerful perceiving capabilities, naive adversarial attacks cannot completely tame SAM. To thoroughly deceive SAM and make it blind to the image, we introduce a frequency-aware optimization strategy, which further enhances the capability of tamper localization. Extensive experimental results demonstrate the effectiveness of our method.

Citations

• ImageSentinel: Protecting Visual Datasets from Unauthorized Retrieval-Augmented Image Generation
• MarkSplatter: Generalizable Watermarking for 3D Gaussian Splatting Model via Splatter Image Structure
• OmniGuard: Hybrid Manipulation Localization via Augmented Versatile Deep Image Watermarking
• Geometry Cloak: Preventing TGS-based 3D Reconstruction from Copyrighted Images
• Transferable Adversarial Attacks on SAM and Its Downstream Models
• Protecting NeRFs' Copyright via Plug-And-Play Watermarking Base Model
• Practical Region-level Attack against Segment Anything Models
• EditGuard: Versatile Image Watermarking for Tamper Localization and Copyright Protection
• Rethinking Image Editing Detection in the Era of Generative AI Revolution
• DRAW: Defending Camera-shooted RAW against Image Manipulation
• SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
• Attack-SAM: Towards Attacking Segment Anything Model With Adversarial Examples
• Micrograph segmentations for DDEVD
• Learning to Immunize Images for Tamper Localization and Self-Recovery
• Hierarchical Text-Conditional Image Generation with CLIP Latents
• From Image to Imuge: Immunized Image Generation
• RWN: Robust Watermarking Network for Image Cropping Localization
• Image Manipulation Detection by Multi-View Multi-Scale Supervision
• On the Robustness of Vision Transformers to Adversarial Examples
• Image Segmentation Using Deep Learning: A Survey
• Towards Deep Learning Models Resistant to Adversarial Attacks
• Universal Adversarial Perturbations Against Semantic Image Segmentation
• Adversarial Examples for Semantic Segmentation and Object Detection
• Adversarial Machine Learning at Scale
• IML-ViT: Benchmarking Image Manipulation Localization by Vision Transformer
• A Threshold Selection Method from Gray-Level Histograms

Related

• ForensicsSAM: Toward Robust and Unified Image Forgery Detection and Localization Resisting to Adversarial Attack
• Zooming In on Fakes: A Novel Dataset for Localized AI-Generated Image Detection with Forgery Amplification Approach
• Black-box Targeted Adversarial Attack on Segment Anything (SAM)
• Research about the Ability of LLM in the Tamper-Detection Area
• Can Vision-Language Models Reason about AI Edits in Images?