Minghan Qin

Tsinghua University, Tsinghua–Berkeley Shenzhen Institute

Papers

Total Citations

H-Index

About

Minghan Qin is a rising researcher in computer vision and 3D scene understanding, with a focus on advancing perception systems for autonomous navigation, augmented reality, and robotics. Their work centers on solving fundamental challenges in object detection, pose estimation, and 3D reconstruction from limited visual data. Qin’s 2024 paper on category-level object detection and pose estimation from stereo images (5 citations) introduces a unified framework that simultaneously detects objects, estimates their 6-DoF poses, and reconstructs their 3D shapes—a critical capability for robots interacting with unknown environments. More recently, their 2025 work on SLGaussian (1 citation) tackles the pressing problem of fast language-guided 3D semantic field learning from sparse views. This method overcomes the inefficiency of traditional multi-view optimization by enabling rapid, accurate scene comprehension with minimal input, directly benefiting applications where data is scarce. Though early in their career, Qin’s contributions demonstrate a clear trajectory toward making 3D perception more robust and efficient, with potential to impact real-world systems in autonomous driving and interactive AR/VR.