OmniDFA: A Unified Framework for Open Set Synthesis Image Detection and Few-Shot Attribution

View on arXiv ← Back to list

Authors: Shiyu Wu, Shuyan Li, Jing Li, Jing Liu, Yequan Wang

Published: 2025-09-30 02:36:40+00:00

AI Summary

OmniDFA is a unified framework proposed for AI-generated image (AIGI) authenticity detection and open-set, few-shot source model attribution. The authors introduce open-set few-shot identification as a new paradigm designed to reliably identify unseen generators using only limited samples. To support this, they construct OmniFake, a large-scale, class-aware synthetic image dataset curating 1.17 million images from 45 distinct generative models.

Abstract

AI-generated image (AIGI) detection and source model attribution remain central challenges in combating deepfake abuses, primarily due to the structural diversity of generative models. Current detection methods are prone to overfitting specific forgery traits, whereas source attribution offers a robust alternative through fine-grained feature discrimination. However, synthetic image attribution remains constrained by the scarcity of large-scale, well-categorized synthetic datasets, limiting its practicality and compatibility with detection systems. In this work, we propose a new paradigm for image attribution called open-set, few-shot source identification. This paradigm is designed to reliably identify unseen generators using only limited samples, making it highly suitable for real-world application. To this end, we introduce OmniDFA (Omni Detector and Few-shot Attributor), a novel framework for AIGI that not only assesses the authenticity of images, but also determines the synthesis origins in a few-shot manner. To facilitate this work, we construct OmniFake, a large class-aware synthetic image dataset that curates $1.17$ M images from $45$ distinct generative models, substantially enriching the foundational resources for research on both AIGI detection and attribution. Experiments demonstrate that OmniDFA exhibits excellent capability in open-set attribution and achieves state-of-the-art generalization performance on AIGI detection. Our dataset and code will be made available.

Key findings

OmniDFA achieved state-of-the-art generalization performance in synthetic image detection, resulting in an average improvement of +5.83% accuracy on the OmniFake dataset compared to previous methods. The framework demonstrates excellent capability in the novel open-set few-shot attribution task, reliably identifying unseen generator categories. Cross-dataset evaluation showed OmniDFA significantly outperformed baselines on challenging zero-shot benchmarks like Chameleon, validating its robustness.

Approach

OmniDFA utilizes a dual-path architecture based on ConvNeXt-Small to capture both fine-grained local and holistic global features of the images. The training leverages supervised contrastive loss to enhance feature discrimination for attribution and incorporates a sphere center loss term for real images to ensure compactness and establish a robust authenticity decision boundary.

Datasets

OmniFake (1.17 M images from 45 distinct generators), GenImage, and Chameleon.

Model(s)

OmniDFA framework built upon ConvNeXt-Small (pretrained on ImageNet) as the feature extractor.

Author countries

China, UK

← Previous