Jiaze Wang

Chinese University of Hong Kong, Shanghai University

Papers

Total Citations

105

H-Index

About

Jiaze Wang is a researcher whose work bridges computer vision and robotics, with a primary focus on category-level 6D object pose estimation—a critical capability for enabling robots to interact with unseen objects in unstructured environments like homes and farms. His most influential contribution, the Cascaded Relation and Recurrent Reconstruction Networks (published in 2021), tackles the fundamental challenge of predicting the precise location and orientation of novel object instances without requiring a pre-scanned 3D model. This work has garnered significant attention, accumulating over 96 citations, and is foundational for advancing robotic manipulation and augmented reality systems. Wang’s approach innovatively recovers instance-specific 3D models in canonical space through a recurrent architecture, setting a new standard for generalization across object categories. Beyond pose estimation, he has applied his engineering expertise to agricultural robotics, designing a flexible end-effector for a tomato harvesting robot (2023). This work demonstrates a practical, systems-level understanding of how perception and manipulation must co-design for real-world tasks. Wang’s research is characterized by its dual impact: advancing core algorithmic challenges in 3D vision while simultaneously deploying those solutions in tangible robotic systems, making his work equally relevant to computer vision theorists and robotics practitioners.