zr-zhang/MLLM-Generated-Image-Detection-Dataset

MLLM-Generated Image Dataset

This dataset contains real and AI-generated image samples organized for binary MLLM-generated image detection.

Paper | Code

Dataset Summary

We construct an MLLM-generated image detection benchmark from GPT Image2 and Nano Banana2. This benchmark covers texture-dominated, structure-dominated, and hybrid-dominated. It is designed to evaluate detector performance under the new challenges introduced by large-scale image generation models.

Usage

from datasets import load_dataset

dataset = load_dataset("zr-zhang/MLLM-Generated-Image-Detection-Dataset")

Licensing and Restrictions

This dataset is made available solely for research on AI-generated content detection, image authenticity analysis, digital forensics, generative model evaluation, and related academic purposes. It must not be used to develop or optimize technologies for fraud, forgery, identity impersonation, fabrication of false evidence, or circumvention of content detection systems.

Citation

If you find this dataset or code useful in your research, please consider citing our work:

@article{zhang2026benchmark,
  title   = {A Benchmark Dataset for MLLM-Generated Image Detection: GPT Image2 & Nano Banana2},
  author  = {Zhang, Zirui and Yu, Yinbo and Guan, Donghai and Tian, Chunwei and Zhang, Daoqiang and Zhu, Qi},
  journal = {arXiv preprint arXiv:2608.01258},
  year    = {2026}
}