Papers
6
Total Citations
106
H-Index
4
About
Jun Yang is a researcher specializing in 3D computer vision, robotic perception, and object pose estimation, with a particular focus on solving real-world challenges in industrial automation and robotic manipulation. His work addresses some of the most demanding problems in the field, including the perception of textureless and highly reflective objects in cluttered environments — conditions that routinely defeat conventional sensing approaches. Yang's most impactful contribution is the ROBI dataset (2021), a multi-view benchmark for reflective objects in robotic bin-picking that has attracted 50 citations, becoming a valuable community resource for researchers tackling glossy, texture-poor industrial parts. Complementing this, his probabilistic multi-view fusion framework for active stereo depth maps demonstrates rigorous thinking about depth map reliability and its downstream effect on 6D pose estimation accuracy. His 2023 work on RGB-only multi-view optimization for textureless object pose estimation further pushes the boundaries by eliminating depth sensor dependency entirely. Beyond bin-picking, Yang has contributed to real-time instance-level 3D reconstruction through InstanceFusion, blending deep learning with SLAM techniques. With citations accumulating across multiple interconnected research threads, Yang has established himself as a thoughtful contributor bridging perception theory and practical robotic deployment.
Research Focus
Key Achievements
Top Papers
- 1ROBI: A Multi-View Dataset for Reflective Objects in Robotic Bin-Picking50 citations · 2021
- 2
- 3
- 4
- 5Point Pair Feature based 6D pose estimation for robotic grasping2 citations · 2020
- 6