Xiaohan Yan   颜小涵

I am currently a Large Model Algorithm Engineer.

I was a master's student majoring in Computer Science and Technology at Tongji University, with my supervisor Gang Wei.

I was an undergraduate student majoring in Computer Science and Technology at Hohai University, where I once served as the captain of the Hohai University ACM Team. My coaches were Xuejie Zhang and Yun Zhu.

My research interests focus on fields such as Multimodal Large Models, 3D Computer Vision, and Reinforcement Learning.

If you find any research interests that we might share, feel free to drop me an email.

 /   /   /   /   / 

中文 / English

• [09/2025] Our paper HOLO is accepted by WACV 2026.


• [03/2025] I graduated from Tongji University with a Master’s degree in Computer Science! 🎓


• [01/2025] Our paper SGGS is accepted by ICASSP 2025.


• [01/2025] Our paper RE0 is accepted by ICRA 2025.

---- show more ----
[07/2024] Our paper AttenPoint is accepted by PRCV 2024.


[04/2024] I will join NIO for a internship!


[03/2024] Our paper ASGMVLP is accepted by MICCAI 2024.


[12/2023] I will join Institute for Al Industry Research (AIR), Tsinghua University for a research internship.🍻!

(* indicates equal contribution)

We propose HOLO, which includes a large-scale scene description dataset and a lightweight 3D-LLM.

Given 3D point clouds and multi-view RGB-D images with poses, RE0 leverages the 3D geometric information, projection relationships and CLIP semantic features for 3D zero-shot instance segmentation.

We use semantic features derived from 2D foundation model to revolutionize the material property optimization for 3DGS.

We combine local and global features of the structures and performance to perform few-shot point cloud semantic segmentation.

A key sub-problem: meta-learning from learning curves is an mature but gradually attention area within the field of meta-learning.

A key sub-problem: meta-learning from learning curves is an mature but gradually attention area within the field of meta-learning.

This competition is about fine-tuning large language models on private datasets.
Our score is 0.905, ranking in the top 3% worldwide and won Silver🏆︎ in the leaderboard.

Scale your device and escape from this geometry storm.
This game reached the "Innovation RK1🥇" and "Theme interpretation RK2🥈" at this game off.