Low-level Visual Perception in the Foundation Model Era

⭐ keywords: Low-level Visual Perception | Multi-Modality Large Language Models | Visual Quality Assessment

What have we published?

Q-Bench: Homepage, Repo, Data, Preprint The first low-level benchmark for foundation models on low-level vision.

What are we working on?

[WIP]Chinese-Q-Bench/质衡: Homepage, Repo The first attempt to test multi-lingual abilities on low-level vision.
[WIP]Q-Instruct: Homepage (TBA), Repo (TBA), Data A large-scale instruction tuning dataset to improve low-level perceptual abilities of foundation models.

Maintained by Teo Wu@Singapore and Zicheng Zhang@Shanghai.