Yiming Qin

Incoming ECE PhD @DukeU, Currently Intern at Impossible Research, Previously Undergrad @PKU1898

[XuDong Wang @XDWang101]

Visual thought shouldn’t be trapped in discrete text space!

Thrilled that CoVT (Chain-of-Visual-Thought) was selected for an oral presentation at #ECCV2026! 🎉

CoVT lets VLMs reason in continuous visual space, beyond discrete text tokens

Code and model: https://t.co/8DoQLzM7Lm

[Yiming Qin @wakals111]

@SuhaoYu1020 Thanks!

[Yiming Qin @wakals111]

🥳 Excited to share that our paper, “Chain-of-Visual-Thought: Teaching VLMs to See and Think Better with Continuous Visual Tokens,” (CoVT) has been selected for an oral presentation at #ECCV2026!

[Yiming Qin @wakals111]

@purshow04 Thank you!

[Yiming Qin @wakals111]

CoVT is only a starting point: we are exploring more possibilities for visual reasoning, with the hope of further advancing breakthroughs in multimodal intelligence.

Project Page: https://t.co/iPxu8ipyLH
Code: https://t.co/d2oXBsROBH
Paper: https://t.co/lNGaO2fLlR

Huge thanks to @jiaxin_ge_ , @XDWang101 , @xkungfu , and all our amazing collaborators for their guidance and support!

[Yiming Qin @wakals111]

Continuous Visual Thinking Works

+5.5 CV-Bench
+14.0 CV-Bench Depth
+3.7 MME-RealWorld
+4.5 HRBench8K

Consistent gains across 10+ benchmarks