| license: cc-by-nc-nd-4.0 | |
| task_categories: | |
| - image-text-to-text | |
| tags: | |
| - omnidirectional-images | |
| - mllm | |
| - benchmark | |
| # ODI-Bench: Can MLLMs Understand Immersive Omnidirectional Environments? | |
| [**Paper**](https://huggingface.co/papers/2510.11549) | [**GitHub**](https://github.com/IntMeGroup/ODI-Bench) | |
| ODI-Bench is a comprehensive benchmark specifically designed for omnidirectional image (ODI) understanding. While multi-modal large language models (MLLMs) excel at conventional 2D images, their ability to comprehend the immersive 360° × 180° environments captured by ODIs is less explored. | |
| The benchmark contains: | |
| - **2,000** high-quality omnidirectional images. | |
| - Over **4,000** manually annotated question-answering (QA) pairs. | |
| - **10** fine-grained tasks covering both general-level and spatial-level ODI understanding. | |
| ## Citation | |
| If you find this work useful, please cite: | |
| ```bibtex | |
| @article{yang2025odi, | |
| title={ODI-Bench: Can MLLMs Understand Immersive Omnidirectional Environments?}, | |
| author={Yang, Liu and Duan, Huiyu and Tao, Ran and Cheng, Juntao and Wu, Sijing and Li, Yunhao and Liu, Jing and Min, Xiongkuo and Zhai, Guangtao}, | |
| journal={arXiv preprint arXiv:2510.11549}, | |
| year={2025} | |
| } | |
| ``` |