A controlled audit on 1,500 images across seven generators finds that training-free detector AUROCs vary by up to 0.38 with backbone choice, preprocessing resolution, and noise sigma, and that score direction is hyperparameter-dependent rather than intrinsic.
Understanding and improv- ing training-free ai-generated image detections with vision foundation models
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
fields
cs.CV 5years
2026 5representative citing papers
DRIFT learns a structured invariance manifold from real images via one-class supervision on decomposed robust and fragile subspaces of a frozen VFM to detect AI-generated images through margin violations.
HydraPrompt uses an Asymmetric Prompt Adapter with fixed real prompts and adaptive fake prompts plus a Conditional Supervised Contrastive loss to achieve SOTA synthetic image detection on benchmarks.
Intermediate layer embedding sensitivity to perturbations distinguishes AI-generated images from real ones, yielding higher AUROC on GenImage and Forensics Small benchmarks than prior methods.
A training-free dual-system framework refines anomaly score ordering on uncertain samples from self-supervised talking head forgery detectors to improve detection performance.
citing papers explorer
-
How Fragile Are Training-Free AI-Generated Image Detectors? A Controlled Audit of Score Direction, Preprocessing, and Compression
A controlled audit on 1,500 images across seven generators finds that training-free detector AUROCs vary by up to 0.38 with backbone choice, preprocessing resolution, and noise sigma, and that score direction is hyperparameter-dependent rather than intrinsic.
-
DRIFT: From Robustness Gaps to Invariance Manifolds for AI-Generated Image Detection
DRIFT learns a structured invariance manifold from real images via one-class supervision on decomposed robust and fragile subspaces of a frozen VFM to detect AI-generated images through margin violations.
-
HydraPrompt: An Adaptive and Asymmetric Framework of Vision-Language Models for Synthetic Image Detection
HydraPrompt uses an Asymmetric Prompt Adapter with fixed real prompts and adaptive fake prompts plus a Conditional Supervised Contrastive loss to achieve SOTA synthetic image detection on benchmarks.
-
Intermediate Representations are Strong AI-Generated Image Detectors
Intermediate layer embedding sensitivity to perturbations distinguishes AI-generated images from real ones, yielding higher AUROC on GenImage and Forensics Small benchmarks than prior methods.
-
Enhancing Self-Supervised Talking Head Forgery Detection via a Training-Free Dual-System Framework
A training-free dual-system framework refines anomaly score ordering on uncertain samples from self-supervised talking head forgery detectors to improve detection performance.