DiffFace-Edit: A Diffusion-Based Facial Dataset for Forgery-Semantic Driven Deepfake Detection Analysis

2026-01-20

AI Summary

This work addresses the limitations of current deepfake detection research, which lacks fine-grained region-level manipulation samples and overlooks the impact of detection-evasion samples—crafted by seamlessly blending real and forged content—on model robustness. To bridge this gap, we introduce DiffFace-Edit, a large-scale dataset comprising over two million diffusion-model-generated facial images, featuring single- and multi-region edits across eight facial areas, and systematically incorporating detection-evasion samples for the first time. Coupled with the In-the-Wild Manipulation Detection Learning (IMDL) framework, we propose a cross-domain evaluation protocol that reveals the significant vulnerability of existing detectors to such evasion samples, thereby establishing a new benchmark and data foundation for robust deepfake detection.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Application Domains: Misinformation & Fake News

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Generative models now produce imperceptible, fine-grained manipulated faces, posing significant privacy risks. However, existing AI-generated face datasets generally lack focus on samples with fine-grained regional manipulations. Furthermore, no researchers have yet studied the real impact of splice attacks, which occur between real and manipulated samples, on detectors. We refer to these as detector-evasive samples. Based on this, we introduce the DiffFace-Edit dataset, which has the following advantages: 1) It contains over two million AI-generated fake images. 2) It features edits across eight facial regions (e.g., eyes, nose) and includes a richer variety of editing combinations, such as single-region and multi-region edits. Additionally, we specifically analyze the impact of detector-evasive samples on detection models. We conduct a comprehensive analysis of the dataset and propose a cross-domain evaluation that combines IMDL methods. Dataset will be available at https://github.com/ywh1093/DiffFace-Edit.

Problem

• deepfake detection
• fine-grained manipulation
• splice attacks
• detector-evasive samples
• facial forgery

Innovation

• Diffusion-based facial editing
• fine-grained manipulation
• detector-evasive samples
• cross-domain evaluation
• deepfake detection