Yiming Qian

Papers

Total Citations

H-Index

About

Yiming Qian is a leading researcher in 3D computer vision and geometric deep learning, with a focus on point cloud processing for robotic perception. Their most-cited work, "Test-Time Adaptation for Point Cloud Upsampling Using Meta-Learning" (2023, 7 citations), addresses a critical bottleneck in real-world 3D sensing: the sparse, non-uniform point clouds generated by affordable scanners. By introducing a meta-learning framework that adapts upsampling models at test time, Qian's approach dramatically improves robustness against domain shifts—a key challenge for deploying autonomous systems in unstructured environments. This contribution bridges the gap between high-performing but brittle architectures and the messy reality of sensor data, enabling more reliable downstream tasks like object recognition and manipulation. Beyond this flagship paper, Qian's research consistently pushes the boundaries of learning from irregular 3D data, with work spanning point cloud completion, registration, and self-supervised representation learning. Their methods have been cited in top venues including CVPR and NeurIPS, and are increasingly adopted in robotics pipelines requiring real-time, adaptive 3D perception. Qian's work exemplifies how meta-learning and test-time adaptation can make deep models practical for real-world sensing.