Mingkun Liu

Papers

Total Citations

H-Index

About

Mingkun Liu is a computer vision researcher whose work centers on 6D pose estimation, a critical technology bridging perception and robotics. His most notable contribution, the NVR-Net (Normal Vector Guided Regression Network), tackles the fundamental challenge of disentangling rotation and translation in monocular 6D pose estimation—a problem that has long plagued existing two-stage methods reliant on Perspective-n-Point (PnP) solvers. By introducing normal vector guidance, Liu's approach achieves more accurate and robust pose predictions, directly addressing the accuracy degeneration inherent in coupled rotation-translation solutions. This work, published in 2023 and already accumulating 19 citations, demonstrates significant early impact in a rapidly evolving field. Liu's research has practical implications for autonomous systems, augmented reality, and industrial robotics, where precise object pose estimation is essential for manipulation and interaction. His innovative framework represents a meaningful step forward in disentangling the geometric complexities of 6D pose, offering a cleaner, more principled alternative to conventional pipelines. As the demand for reliable visual perception in robotics grows, Liu's contributions position him as a rising voice in advancing the theoretical and applied frontiers of computer vision.