Qinglin “Quinn” Yang (杨青林)
I am a Ph.D. student in the College of Artificial Intelligence at Zhejiang University, advised by Prof. Ruizhen Hu.
Before starting my Ph.D., I pursued master’s studies at Zhejiang University’s School of Software Technology, advised by Prof. Sida Peng.
I am interested in the nature of intelligence, with a focus on exploring recursive self-improvement. My earlier research spans mesh generation, metric depth estimation, and dynamic scene reconstruction.
What I work on
Looking ahead, I plan to explore recursive self-improvement (RSI) and what it can reveal about the nature of intelligence. I am particularly interested in learning and reasoning, self-reference, and the foundations of intelligence.
Education
Zhejiang University · Ph.D. Student
College of Artificial Intelligence
Advisor: Prof. Ruizhen Hu
Zhejiang University · Master’s Studies
Electronic Information, School of Software Technology
Advisor: Prof. Sida Peng
Transferred to the Ph.D. program in September 2026.
Wuhan University · Bachelor’s Degree
Spatial Information and Digital Technology
School of Remote Sensing and Information Engineering
Publications
* Equal contribution.
Towards Artistic Mesh Generation via Part-wise Autoregressive Modeling
Journal of Computer-Aided Design & Computer Graphics (JCAD), 2026 · Accepted
MeshPack generates high-resolution, part-aware meshes through part-wise autoregressive modeling and cross-part information passing, trained on the FurniSet3D furniture dataset.
Towards Depth Foundation Models: Recent Trends in Vision-Based Depth Estimation
Computational Visual Media, 12(2): 243–271, 2026
This work summarizes key architectures and training strategies for robust depth estimation, outlining the path toward depth foundation models and practical applications.
Hierarchy UGP: Hierarchy Unified Gaussian Primitive for Large-Scale Dynamic Scene Reconstruction
IEEE/CVF International Conference on Computer Vision (ICCV), pp. 26252–26262, 2025
Combining hierarchy and unified Gaussian primitives to efficiently reconstruct large-scale dynamic street scenes.
Manuscript
PromptDepthAnything++: Accurate 4K Metric Depth Estimation via Pattern-Agnostic Prompting
Manuscript submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence
Pattern-agnostic sparse-point prompting for accurate, high-resolution metric depth estimation.
Internships
TeleAI
Intern · World Action Model
Alibaba (Taotian Group), Meta Team
Algorithm Intern · 3D Generation
Lightwheel
Algorithm Intern · Street-Scene Reconstruction
LuoJiaNet Group, Wuhan University
Algorithm Intern · 3D Reconstruction
For more details, view my CV.