ML Security PapersSemantic Discrepancy-aware Detector for Image Forgery Identification
Ziye Wang 1, Minghang Yu 1, Chunyan Xu 1, Zhen Cui 2
Published on arXiv
2508.12341
Output Integrity Attack
OWASP ML Top 10 — ML09
Key Finding
SDD achieves superior image forgery detection results compared to existing methods on two standard image forgery benchmarks
SDD (Semantic Discrepancy-aware Detector)
Novel technique introduced
With the rapid advancement of image generation techniques, robust forgery detection has become increasingly imperative to ensure the trustworthiness of digital media. Recent research indicates that the learned semantic concepts of pre-trained models are critical for identifying fake images. However, the misalignment between the forgery and semantic concept spaces hinders the model's forgery detection performance. To address this problem, we propose a novel Semantic Discrepancy-aware Detector (SDD) that leverages reconstruction learning to align the two spaces at a fine-grained visual level. By exploiting the conceptual knowledge embedded in the pre-trained vision language model, we specifically design a semantic token sampling module to mitigate the space shifts caused by features irrelevant to both forgery traces and semantic concepts. A concept-level forgery discrepancy learning module, built upon a visual reconstruction paradigm, is proposed to strengthen the interaction between visual semantic concepts and forgery traces, effectively capturing discrepancies under the concepts' guidance. Finally, the low-level forgery feature enhancemer integrates the learned concept level forgery discrepancies to minimize redundant forgery information. Experiments conducted on two standard image forgery datasets demonstrate the efficacy of the proposed SDD, which achieves superior results compared to existing methods. The code is available at https://github.com/wzy1111111/SSD.
Key Contributions
- Semantic token sampling module to mitigate feature space misalignment between forgery traces and semantic concepts in pre-trained vision-language models
- Concept-level forgery discrepancy learning module built on a visual reconstruction paradigm to capture forgery discrepancies guided by semantic concepts
- Low-level forgery feature enhancer that integrates concept-level discrepancies to reduce redundant forgery information
🛡️ Threat Analysis
The paper's primary contribution is a novel AI-generated image detection architecture — directly fitting the ML09 category of AI-generated content detection and output integrity. It proposes forensic components (semantic token sampling, concept-level forgery discrepancy learning) specifically to identify fake/forged images produced by modern generative techniques.