Reproduce “Look Before You Judge: Training-Free Region Mining for Grounded and Explainable Deepfake Detection” 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
Chia-Ling Chen, Yu-Ting Ta, Jian-Yu Jiang-Lin, Tai-Ming Huang, Ling Lo, Po-Ching Chen, Yan-Tsung Wang, Pei-Heng Li and 3 more
The authors introduce a training-free framework for explainable deepfake detection that mines image-specific regions, inspects them separately, then checks observations against the whole image. On TriDF, with InternVL-3.5-8B, authenticity accuracy rises from 0.4176 to 0.5458 versus vanilla inference. The proposed regions are inspection candidates, not causal explanations, and the method requires white-box decoder attention.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.