A Survey on Generalizable Diffusion-generated Image Detection

Qijie Xu, Defang Chen, Jiawei Chen, Siwei Lyu, Can Wang

The repository is based on our recently released survey Recent Advances on Generalizable Diffusion-generated Image Detection.

Abstract

The rise of diffusion models has significantly improved the fidelity and diversity of generated images. With numerous benefits, these advancements also introduce new risks. Diffusion models can be exploited to create high-quality Deepfake images, which poses challenges for image authenticity verification. In recent years, research on generalizable diffusion-generated image detection has grown rapidly. However, a comprehensive review of this topic is still lacking. To bridge this gap, we present a systematic survey of recent advances and classify them into two main categories: (1) data-driven detection and (2) feature-driven detection. Existing detection methods are further classified into six fine-grained categories based on their underlying principles. Finally, we identify several open challenges and envision some future directions, with the hope of inspiring more research work on this important topic.

BibTex

If you find our survey useful for your research project, please consider citing our paper:

@article{xu2025recent,
  title={Recent Advances on Generalizable Diffusion-generated Image Detection},
  author={Xu, Qijie and Chen, Defang and Chen, Jiawei and Lyu, Siwei and Wang, Can},
  journal={arXiv preprint arXiv:2502.19716},
  year={2025}
}

Contents

• Overview
• Papers
  • Data-driven Detection
  • Feature-driven Detection
    • Perceptible Image Features
    • Imperceptible Image Features
    • Features Beyonds images

Overview

Papers

Data-driven Detection

We classify data-driven detection methods into three categories: (1) advanced model architectures, (2) reduced dataset bias, and (3) improved training objectives, denoted as "Architecture", "Dataset" and "Objective", respectively.

• Towards Universal Fake Image Detectors that Generalize Across Generative Models

  • CVPR 2023
  • https://arxiv.org/abs/2302.10174
  • Utkarsh Ojha, Yuheng Li, Yong Jae Lee
  • Architecture

• GenDet: Towards Good Generalizations for AI-Generated Image Detection

  • https://arxiv.org/abs/2312.08880
  • Mingjian Zhu, Hanting Chen, Mouxiao Huang, Wei Li, Hailin Hu, Jie Hu, Yunhe Wang
  • Objective

• Forgery-aware Adaptive Transformer for Generalizable Synthetic Image Detection

  • CVPR 2024
  • https://arxiv.org/abs/2312.16649
  • Huan Liu, Zichang Tan, Chuangchuang Tan, Yunchao Wei, Yao Zhao, Jingdong Wang
  • Architecture + Objective

• Frequency Masking for Universal Deepfake Detection

  • ICASSP 2024
  • https://arxiv.org/abs/2401.06506
  • Chandler Timm Doloriel, Ngai-Man Cheung
  • Dataset

• DRCT: Diffusion Reconstruction Contrastive Training towards Universal Detection of Diffusion Generated Images

  • ICML 2024
  • https://proceedings.mlr.press/v235/chen24ay.html
  • Baoying Chen, Jishen Zeng, Jianquan Yang, Rui Yang
  • Dataset + Objective

• CLIPping the Deception: Adapting Vision-Language Models for Universal Deepfake Detection

  • ICMR 2024
  • https://arxiv.org/abs/2402.12927
  • Sohail Ahmed Khan, Duc-Tien Dang-Nguyen
  • Objective

• Leveraging Representations from Intermediate Encoder-blocks for Synthetic Image Detection

  • ECCV 2024
  • https://arxiv.org/abs/2402.19091
  • Christos Koutlis, Symeon Papadopoulos
  • Architecture

• Zero-Shot Detection of AI-Generated Images

  • ECCV 2024
  • https://arxiv.org/abs/2409.15875
  • Davide Cozzolino, Giovanni Poggi, Matthias Nießner, Luisa Verdoliva
  • Objective

• Mixture of Low-rank Experts for Transferable AI-Generated Image Detection

  • https://arxiv.org/abs/2404.04883
  • Zihan Liu, Hanyi Wang, Yaoyu Kang, Shilin Wang
  • Architecture

• SemGIR: Semantic-Guided Image Regeneration Based Method for AI-generated Image Detection and Attribution

  • ACM MM 2024
  • https://dl.acm.org/doi/abs/10.1145/3664647.3680776
  • Xiao Yu, Kejiang Chen, Kai Zeng, Han Fang, Zijin Yang, Xiuwei Shang, Yuang Qi, Weiming Zhang, Nenghai Yu
  • Dataset

• Aligned Datasets Improve Detection of Latent Diffusion-Generated Images

  • ICLR 2025
  • https://arxiv.org/abs/2410.11835v3
  • Anirudh Sundara Rajan, Utkarsh Ojha, Jedidiah Schloesser, Yong Jae Lee
  • Dataset

Feature-driven Detection

Perceptible Image Features

• DE-FAKE: Detection and Attribution of Fake Images Generated by Text-to-Image Generation Models

  • CCS 2023
  • https://arxiv.org/abs/2210.06998
  • Zeyang Sha, Zheng Li, Ning Yu, Yang Zhang

• Shadows Don't Lie and Lines Can't Bend! Generative Models don't know Projective Geometry...for now

  • CVPR 2024
  • https://arxiv.org/abs/2311.17138
  • Ayush Sarkar, Hanlin Mai, Amitabh Mahapatra, Svetlana Lazebnik, D.A. Forsyth, Anand Bhattad

Imperceptible Image Features

• PatchCraft: Exploring Texture Patch for Efficient AI-generated Image Detection

  • https://arxiv.org/abs/2311.12397
  • Nan Zhong, Yiran Xu, Sheng Li, Zhenxing Qian, Xinpeng Zhang
  • High-frequency noise

• Rethinking the Up-Sampling Operations in CNN-based Generative Network for Generalizable Deepfake Detection

  • CVPR 2024
  • https://arxiv.org/abs/2312.10461
  • Chuangchuang Tan, Huan Liu, Yao Zhao, Shikui Wei, Guanghua Gu, Ping Liu, Yunchao Wei
  • Local correlations

• A Single Simple Patch is All You Need for AI-generated Image Detection

  • https://arxiv.org/abs/2402.01123
  • Jiaxuan Chen, Jieteng Yao, Li Niu
  • High-frequency noise

• Improving Synthetic Image Detection Towards Generalization: An Image Transformation Perspective

  • https://arxiv.org/abs/2408.06741
  • Ouxiang Li, Jiayin Cai, Yanbin Hao, Xiaolong Jiang, Yao Hu, Fuli Feng
  • Local correlations

• A Sanity Check for AI-generated Image Detection

  • ICLR 2025
  • https://arxiv.org/abs/2406.19435
  • Shilin Yan, Ouxiang Li, Jiayin Cai, Yanbin Hao, Xiaolong Jiang, Yao Hu, Weidi Xie
  • High-frequency noise

• Leveraging Natural Frequency Deviation for Diffusion-Generated Image Detection

  • https://openreview.net/forum?id=fPBExgC1m9
  • Daichi Zhang, Tong Zhang, Shiming Ge, Sabine Susstrunk
  • Frequency domain

Features Beyonds images

• DIRE for Diffusion-Generated Image Detection

  • ICCV 2023
  • https://arxiv.org/abs/2303.09295
  • Zhendong Wang, Jianmin Bao, Wengang Zhou, Weilun Wang, Hezhen Hu, Hong Chen, Houqiang Li

• AEROBLADE: Training-Free Detection of Latent Diffusion Images Using Autoencoder Reconstruction Error

  • CVPR 2024
  • https://arxiv.org/abs/2401.17879
  • Jonas Ricker, Denis Lukovnikov, Asja Fischer

• FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion

  • CVPR 2024
  • https://arxiv.org/abs/2406.08603
  • George Cazenavette, Avneesh Sud, Thomas Leung, Ben Usman

• LaRE$^2$: Latent Reconstruction Error Based Method for Diffusion-Generated Image Detection

  • CVPR 2024
  • https://arxiv.org/abs/2403.17465
  • Yunpeng Luo, Junlong Du, Ke Yan, Shouhong Ding

• ZeroFake: Zero-Shot Detection of Fake Images Generated and Edited by Text-to-Image Generation Models

  • CCS 2024
  • https://dl.acm.org/doi/abs/10.1145/3658644.3690297
  • Zeyang Sha, Yicong Tan, Mingjie Li, Michael Backes, Yang Zhang

• Manifold Induced Biases for Zero-shot and Few-shot Detection of Generated Images

  • ICLR 2025
  • https://openreview.net/forum?id=7gGl6HB5Zd
  • Jonathan Brokman, Amit Giloni, Omer Hofman, Roman Vainshtein, Hisashi Kojima, Guy Gilboa