Apr 12, 2024 · This work introduces Mixture-of-Experts modules for Face Forgery Detection (MoE-FFD), a generalized yet parameter-efficient ViT-based approach.

Jun 10, 2024 · Face Forgery Detection (FFD), or Deepfake detection, aims to determine whether a digital face is real or fake. Due to different face synthesis ...

ViT-based face forgery detection approach overcoming generalizability and computational issues.

MoE in Visual Domain ; MLLMs, MoE-FFD: Mixture of Experts for Generalized and Parameter-Efficient Face Forgery Detection, arXiv, 2024 ; MoE-LLaVA, MoE-LLaVA: ...

MoE-FFD: Mixture of Experts for Generalized and Parameter-Efficient Face Forgery Detection · 1 code implementation • 12 Apr 2024 • Chenqi Kong, Anwei Luo, ...

Topics · Cross-Manipulation · Low-Rank Adaptation · Forgery Faces · Fake News · Training Phase · Vision Transformer ...

Moe-ffd: Mixture of experts for generalized and parameter-efficient face forgery detection. C Kong, A Luo, P Bao, Y Yu, H Li, Z Zheng, S Wang, AC Kot. arXiv ...

MoE-FFD: Mixture of Experts for Generalized and Parameter-Efficient Face Forgery Detection · 1 code implementation • 12 Apr 2024 • Chenqi Kong, Anwei Luo, ...

MoE-FFD improves face forgery detection with specialized experts for different types of forgeries. The use of adapters contributes to parameter efficiency.

To solve the mentioned problems, the paper proposes a novel generalized residual Federated learning for face Forgery detection (FedForgery). The designed ...