Ming-Yu Liu
Papers
1
Total Citations
43
H-Index
1
About
Ming-Yu Liu is a leading researcher in computer vision and machine learning, with a focus on domain adaptation, generative models, and synthetic data for autonomous systems. His most-cited work, "Domain Stylization: A Fast Covariance Matching Framework Towards Domain Adaptation" (2020, 43 citations), addresses a critical bottleneck in robotics and autonomous driving: the domain gap between synthetic and real-world data. Liu introduced a fast covariance matching technique that stylizes computer graphics rendered images to better align with real-world distributions, enabling models trained purely on synthetic data to generalize effectively. This contribution has been instrumental in reducing reliance on costly real-world labeled datasets, accelerating progress in simulation-to-real transfer. Beyond this, Liu’s broader research spans generative adversarial networks, image-to-image translation, and self-supervised learning, with his papers collectively amassing thousands of citations. His work is widely recognized for bridging theory and practical deployment, influencing both academic benchmarks and industry applications in autonomous driving and robotics. Liu’s ability to tackle fundamental challenges in domain adaptation has made him a key figure in advancing robust, scalable vision systems.
Research Focus
Key Achievements
Top Papers
- 1