HiDF: A Human-Indistinguishable Deepfake Dataset

Chaewon Kang, Seoyoon Jeong, jonghyun lee, Daejin Choi, Simon S. Woo, Jinyoung Han

Keywords: Deepfake, Human-Indistinguishable, AI, Multimodal

TL;DR: In this paper, we introduce HiDF, a novel high-quality and human-indistinguishable deepfake dataset, which comprises 30K deepfake images and 4K deepfake videos.

Abstract: The rapid development and prevalence of generative AI has made it easy for people to create high-quality deepfake images and videos, but their abuses also have been exponentially increased. To mitigate potential social disruption, it is crucial to quickly detect authenticity of each deepfake content hidden in a sea of information. While researchers have worked on developing deep learning-based methods, the deepfake datasets utilized in these studies are far from the real world in terms of their qualities; most of the popular deepfake datasets are human distinguishable. To address this problem, we present a novel deepfake dataset, HiDF, a high-quality and human-indistinguishable deepfake dataset consisting of 30 K images and 4 K videos. HiDF is a meticulously curated dataset that includes diverse subjects, which has been undergone rigorous quality checks. Comparison on the quality between HiDF and existing deepfake datasets demonstrates that HiDF is human-indistinguishable, hence it can be used as a valuable benchmark dataset for deepfake detection tasks. Data and code (https://github.will.be.provided) are publicly available for future deepfake detection research.

Primary Area: datasets and benchmarks

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 7654