Data Release

We release our datasets in huggingface datasets.

A1: LLVisionQA (Perception)

Specifically, for methods to improve their low-level ability, we split the datasets into two-halves: the dev subset, and the test subset, with release information shown as follows:

The respective images are in the images.tar under the huggingface repository, as well as the json files for dev and test subsets.

A2: LLDescribe (Description)

Only images are released in images.tar. Please refer here to download them.

Contact

Please contact any of the first authors of this paper for queries.

Haoning Wu, haoning001@e.ntu.edu.sg, @teowu
Zicheng Zhang, zzc1998@sjtu.edu.cn, @zzc-1998
Erli Zhang, ezhang005@e.ntu.edu.sg, @ZhangErliCarl

Citation

If you find our work interesting, please feel free to cite our paper:

@article{wu2023qbench,
    title={Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision},
    author={Wu, Haoning and Zhang, Zicheng and Zhang, Erli and Chen, Chaofeng and Liao, Liang and Wang, Annan and Li, Chunyi and Sun, Wenxiu and Yan, Qiong and Zhai, Guangtao and Lin, Weisi},
    year={2023},
}