Zeming Fan
Papers
2
Total Citations
16
H-Index
2
About
Dr. Zeming Fan is a rising innovator at the intersection of computer vision, deep learning, and agricultural robotics. His research focuses on developing lightweight, efficient neural network architectures for 3D reconstruction, particularly applied to orchard environments. Fan’s major contribution lies in pioneering mobile transformer-based convolutional networks that dramatically reduce computational and memory demands while maintaining high reconstruction accuracy. His flagship work, MT-MVSNet, introduces a mobile transformer framework for Multi-view Stereo, achieving state-of-the-art results in reconstructing complex, fine-detail structures like fruit tree branches—a notoriously challenging task due to occlusion and irregular geometry. With 10 citations in its first year, this work has quickly become a reference point for efficient 3D vision in agriculture. His follow-up model, ET-PatchNet, further advances low-memory, high-efficiency stereo matching, garnering 6 citations in 2025. By enabling real-time, high-fidelity 3D mapping of orchard canopies on edge devices, Fan’s research directly supports precision agriculture, automated pruning, and yield estimation. His work exemplifies how tailored deep learning can bridge the gap between state-of-the-art computer vision and practical, resource-constrained field applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2