Gu Wang

Papers

Total Citations

819

H-Index

About

Gu Wang is a leading researcher in computer vision, specializing in 6D object pose estimation and its applications in robotics and augmented reality. His most influential contribution is the development of DeepIM, a deep iterative matching framework that revolutionized 6D pose estimation by enabling precise alignment of 3D object models to images through iterative refinement. This work, published in 2018 and 2019, has garnered over 780 citations combined, cementing its status as a foundational method in the field. DeepIM’s ability to handle large pose variations and occlusions set new benchmarks, inspiring subsequent research in robust object tracking and manipulation. Wang further advanced the domain with CPS++, a self-supervised learning approach that extends 6D pose and shape estimation to class-level recognition from monocular images. This work addresses the critical challenge of scaling pose estimation to hundreds of object instances—a necessity for deploying robots in unstructured environments. By reducing reliance on expensive 3D annotations, CPS++ paves the way for practical, real-world applications. Wang’s research continues to bridge the gap between laboratory performance and industrial deployment, making him a pivotal figure in modern computer vision.