Forensic Reserve: Eliciting Latent Knowledge for Image Forgery Detection
To address this limitation, we propose Reserve-Guided Elicitation (RGE), a framework that treats sparse, origin-sensitive internal components in pretrained models as a forensic reserve and translates their localization into structural constraints for lightweight adaptation.
ProofPaper ↗
Key points
- As generated images become increasingly realistic, reliable forgery detection is essential for maintaining trust in visual information.
- However, existing methods primarily rely on task-specific supervision to adapt vision foundation model representations, without fully exploiting internal forensic knowledge to guide detection.
- Specifically, we first use the Forensic Lens (F-lens) to decompose activations across layers and token groups into independent components and globally screen them by their response differences between real and generated images, identifying reserve sites and directions.
- Furthermore, RGE consistently improves over the corresponding frozen detectors across eight encoders spanning self-supervised and vision-language pretraining, eliciting a latent forensic capacity broadly shared across pretrained vision models.
Sources (1)
- [1]Forensic Reserve: Eliciting Latent Knowledge for Image Forgery DetectionarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 04:33 PM
To address this limitation, we propose Reserve-Guided Elicitation (RGE), a framework that treats sparse, origin-sensitive internal components in pretrained models as a forensic reserve and translates their localization into structural constraints for lightweight adaptation.
As generated images become increasingly realistic, reliable forgery detection is essential for maintaining trust in visual information.
Extractive summary: sentences quoted from the sources.