Ming‐Ming Cheng

Nankai University, University of Oxford, Chinese Academy of Sciences

Papers

Total Citations

165

H-Index

About

Ming-Ming Cheng is a leading researcher in computer vision and interactive 3D scene understanding, best known for pioneering real-time, user-guided semantic segmentation of 3D environments. His most impactful work, the SemanticPaint system (2015), has garnered over 160 combined citations for its two core publications. This breakthrough allows users to simultaneously scan and interactively label 3D spaces by simply touching objects, enabling the system to continuously learn and update its segmentation model on the fly. Cheng’s contributions fundamentally shifted the paradigm from offline, batch processing to online, human-in-the-loop 3D understanding, making semantic labeling accessible and practical for augmented reality, robotics, and human-computer interaction. His work demonstrates how intuitive, real-time interaction can bridge the gap between raw sensor data and meaningful scene interpretation. By empowering users to teach their devices as they explore, Cheng has opened new avenues for adaptive and collaborative AI systems. His research continues to inspire advances in interactive machine learning, with SemanticPaint remaining a foundational reference for anyone working on dynamic, user-centric 3D vision.