Reproduce “DeepfakeGenome: Toward Next-Generation Deepfake Attribution” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shunli Wang, Xinyu Zhou, Yandan Zhao, Wenbin Liang, Taiping Yao, Ke-Yue Zhang, Zhiyuan Yan, Shouhong Ding and 1 more
The authors introduce DeepfakeGenome, a benchmark containing 100 algorithms and 2 million images across four forgery categories. It evaluates real/fake detection, attribution, and retrieval for cross-domain, cross-algorithm testing. On Protocol 4, fine-tuned CLIP-ViT achieves Ret-mAP 0.5507, above DINOv2-ViT-FFT’s 0.3618; this is a benchmark result, not evidence of deployed defense performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.