DeepFake-o-meter

Task: Face Forgery Detection

Analysis Date: 2025-11-19 05:53:38.724496

Information:
Resolution: 1024 x 608 | Codec: JPEG | File Size: 66.39KB

Analytic Name: CFM

Description: A framework that, through prior-knowledge-agnostic augmentations, fine-grained triplet relation learning, and a progressive learning controller, systematically mines and focuses on the most discriminative forgery clues to achieve state-of-the-art robustness and generalization in face forgery detection.

Reference: https://arxiv.org/abs/2304.12489

Code Link: https://github.com/LoveSiameseCat/CFM

🧪 Fake Probability: 41.57%

▲

41.57%

AI-generated Likelihood

Scope: Human face images and videos

Result Details