Few-shot forgery detection via Guided Adversarial Interpolation

Haonan Qiu, Siyu Chen, Bei Gan, Kun Wang, Huafeng Shi, Jing Shao, Ziwei Liu

Abstract: Highlights•We build four datasets and form a few-shot forgery detection benchmark.•We propose a concise yet effective framework, Guided Adversarial Interpolation (GAI).•Experiments demonstrate that GAI achieves SOTA performance on established benchmark.