🧠 R2I-Bench: Benchmarking Reasoning-Driven Text-to-Image Generation

🌟Official repository for the paper: ⭐️R2I-Bench: Benchmarking Reasoning-Driven Text-to-Image Generation⭐️

---

🌈 Overview: Reasoning in Image Generation

We introduce Logo R2I-Bench , a comprehensive benchmark designed to assess the reasoning capabilities of text-to-image (T2I) generation models.

---

🚀 Get Started

📦 Installation

Clone the repository:

git clone https://github.com/PLUM-Lab/R2I-Bench
cd R2I-Bench

💡 Make sure you have Python ≥ 3.8 and install dependencies listed in requirements.txt.

---

🖼️ Generation

Run the generation script:

python generation.py

This will produce model-generated images based on reasoning-oriented prompts.

---

📊 Evaluation

To evaluate generated results:

python evaluation.py

🎈 Contact

2252538@tongji.edu.cn

---

📌 Citation

🎀 If you find this repository helpful in your research or applications, please consider citing:

@inproceedings{chen2025r2i,
  title={R2i-bench: Benchmarking reasoning-driven text-to-image generation},
  author={Chen, Kaijie and Lin, Zihao and Xu, Zhiyang and Shen, Ying and Yao, Yuguang and Rimchala, Joy and Zhang, Jiaxin and Huang, Lifu},
  booktitle={Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  pages={12606--12641},
  year={2025}
}