Papers
7
Total Citations
202
H-Index
5
About
Minglun Gong is a leading researcher in computer vision and graphics, with a focus on 3D reconstruction, real-time stereo matching, and autonomous scanning. His most influential work, "Quality-driven Poisson-guided autoscanning" (78 citations), revolutionized 3D model acquisition by prioritizing scan quality over coverage, introducing a Poisson-guided method that ensures high-fidelity surface reconstruction. In real-time stereo vision, Gong's 2006 paper "How Far Can We Go with Local Optimization in Real-Time Stereo Matching" (58 citations) pushed the boundaries of local optimization, demonstrating that high-accuracy disparity maps are achievable under strict time constraints—critical for robot navigation and augmented reality. He further advanced temporal consistency in stereo estimation (45 citations), enabling smoother, more reliable video-rate depth estimation. Gong's recent work, "Neural Packing: from Visual Sensing to Reinforcement Learning" (2023), bridges computer vision and robotics, introducing a full pipeline for 3D transport-and-packing using RGBD sensing and reinforcement learning. With a career spanning foundational stereo algorithms to cutting-edge neural packing, Gong's contributions have garnered over 200 citations, solidifying his impact on both theoretical and applied computer vision.
Research Focus
Key Achievements
Top Papers
- 1Quality-driven poisson-guided autoscanning78 citations · 2014
- 2How Far Can We Go with Local Optimization in Real-Time Stereo Matching58 citations · 2006
- 3Enforcing Temporal Consistency in Real-Time Stereo Estimation45 citations · 2006
- 4Neural Packing: from Visual Sensing to Reinforcement Learning8 citations · 2023
- 5
- 6Adaptive Framework for Robust Visual Tracking5 citations · 2018
- 73D Visual Homing for Commodity UAVs2 citations · 2018