Papers
3
Total Citations
266
H-Index
3
About
Shao-Yuan Lo is a researcher at the forefront of efficient computer vision, with a primary focus on real-time semantic segmentation and domain-adaptive depth estimation. His most influential contribution is the development of Efficient Dense Modules of Asymmetric Convolution (EDANet), a groundbreaking architecture that redefined the balance between speed and accuracy for real-time semantic segmentation. This work, which has garnered over 228 citations, introduced a novel approach using dense connections and asymmetric convolutions to achieve high-performance inference without sacrificing computational efficiency—a critical advancement for applications in autonomous driving and robotics. Lo further extended his impact into 3D perception with his work on domain adaptive monocular depth estimation, where he proposed learning feature decomposition to bridge the gap between synthetic and real-world data. His research directly addresses the practical challenges of deploying deep learning in resource-constrained, real-world environments, making his contributions highly valued by both academia and industry. Through his innovative architectures and domain adaptation techniques, Lo has established himself as a key figure in enabling efficient, robust visual perception for autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3