Yunfan Ye

National University of Defense Technology, Hunan University

Papers

Total Citations

H-Index

About

Yunfan Ye is a researcher advancing the frontiers of computer vision and robotics, with a primary focus on template matching and dexterous manipulation. His most cited work, "Learning accurate template matching with differentiable coarse-to-fine correspondence refinement" (2024, 17 citations), tackles a longstanding challenge in manufacturing and robotic grasping: accurately estimating object poses from images. By introducing a differentiable coarse-to-fine refinement pipeline, Ye's method significantly improves robustness where traditional approaches fail, enabling precise part localization for downstream tasks like assembly and pick-and-place. This contribution directly addresses real-world industrial needs, bridging the gap between classical vision techniques and modern deep learning. Additionally, in "Learning Cross-Hand Policies of High-DOF Reaching and Grasping" (2024), Ye explores how to generalize manipulation skills across different robotic hand morphologies, a critical step toward adaptable, high-degree-of-freedom robotic systems. His work demonstrates a rare ability to combine theoretical rigor with practical deployment, making him a rising voice in applied vision and robotics. For students and researchers, Ye's research offers a clear path from foundational computer vision problems to cutting-edge robotic applications, highlighting the power of differentiable methods in solving long-standing industrial challenges.