DFBench
收藏DFBench 数据集概述
数据集基本信息
- 名称: DFBench
- 用途: 评估大型多模态模型在深度伪造图像检测方面的能力
- 相关论文: DFBench: Benchmarking Deepfake Image Detection Capability of Large Multimodal Models
- 数据集地址: Hugging Face 数据集
数据集内容
- 数据文件:
img_train_shuffled.jsonimg_test.jsonimg_train_shuffled.jsonlimg_test.jsonl
下载方式
-
Linux: bash export HF_ENDPOINT=https://hf-mirror.com huggingface-cli download IntMeGroup/DFBench --repo-type dataset --local-dir ./DFBench
-
Windows Powershell: bash $env:HF_ENDPOINT = "https://hf-mirror.com" huggingface-cli download IntMeGroup/DFBench --repo-type dataset --local-dir ./DFBench
数据集特征
- 特征分布:
- 真实图像(无失真)
- 真实图像(有失真)
- AI编辑图像
- AI生成图像
相关模型
- Qwen2.5-VL-7B-Instruct
- InternVL2_5-8B
- InternVL3-9B
训练与评估
- 训练脚本:
train.sh - 评估脚本:
eval_deepfake.sh - 结果计算:
logit_calculation.py,process_results.py
可视化
- 特征分布图:
feature_distribution.py,plot_features.py
引用
bibtex @misc{wang2025dfbenchbenchmarkingdeepfakeimage, title={DFBench: Benchmarking Deepfake Image Detection Capability of Large Multimodal Models}, author={Jiarui Wang and Huiyu Duan and Juntong Wang and Ziheng Jia and Woo Yi Yang and Xiaorong Zhu and Yu Zhao and Jiaying Qian and Yuke Xing and Guangtao Zhai and Xiongkuo Min}, year={2025}, eprint={2506.03007}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2506.03007}, }
联系方式
- 邮箱: wangjiarui@sjtu.edu.cn




