Zhengfa Liang
Papers
1
Total Citations
9
H-Index
1
About
Zhengfa Liang is a leading researcher in computer vision, with a primary focus on stereo matching for robotic navigation and autonomous systems. His work tackles the critical challenge of balancing accuracy and computational efficiency in depth estimation. Liang’s most influential contribution is the **Multi-Scale Cost Volumes Cascade Network (MSCVNet)**, a novel deep learning architecture that cascades cost volumes at multiple scales to achieve high-precision stereo matching without the prohibitive computational cost typical of CNN-based methods. This approach directly addresses the limitations of traditional algorithms, which suffer from low accuracy, and heavy neural networks, which are too slow for real-time applications. By intelligently fusing cost volumes, MSCVNet delivers state-of-the-art performance while maintaining practical runtime, making it a key reference for researchers developing efficient 3D perception systems. With 9 citations on his landmark 2021 paper, Liang’s work is gaining traction as a foundational solution for robots that must navigate complex environments quickly and reliably. His research continues to push the boundaries of efficient, accurate depth sensing for embodied AI.
Research Focus
Key Achievements
Top Papers
- 1Multi-Scale Cost Volumes Cascade Network for Stereo Matching9 citations · 2021