DF-LLaVA: Unlocking MLLM's potential for Synthetic Image Detection via Prompt-Guided Knowledge Injection
This addresses the challenge of accurately detecting synthetic images with interpretable results for applications in media verification and security, representing a strong specific gain.
The paper tackled the problem of synthetic image detection by proposing DF-LLaVA, a framework that enhances MLLMs to achieve higher accuracy than expert models while maintaining interpretability, with results showing outstanding detection accuracy.
With the increasing prevalence of synthetic images, evaluating image authenticity and locating forgeries accurately while maintaining human interpretability remains a challenging task. Existing detection models primarily focus on simple authenticity classification, ultimately providing only a forgery probability or binary judgment, which offers limited explanatory insights into image authenticity. Moreover, while MLLM-based detection methods can provide more interpretable results, they still lag behind expert models in terms of pure authenticity classification accuracy. To address this, we propose DF-LLaVA, a simple yet effective framework that unlocks the intrinsic discrimination potential of MLLMs. Our approach first extracts latent knowledge from MLLMs and then injects it into training via prompts. This framework allows LLaVA to achieve outstanding detection accuracy exceeding expert models while still maintaining the interpretability offered by MLLMs. Extensive experiments confirm the superiority of our DF-LLaVA, achieving both high accuracy and explainability in synthetic image detection. Code is available online at: https://github.com/Eliot-Shen/DF-LLaVA.