Multi-definition Deepfake detection via semantics reduction and cross-domain training

Published: 01 Jan 2025, Last Modified: 13 May 2025Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Analyzed the challenges in Deepfake detection under complex real-world scenarios.•Reduce high-level semantics learning in Deepfake detection to enhance robustness.•Adversarial learning & distillation for improved generalization capability.•Achieves SOTA performance; also effective on Face Anti-Spoofing.
Loading