Liyuan Zhu

Papers

Total Citations

H-Index

About

Liyuan Zhu is a researcher at the forefront of applying self-supervised learning to environmental monitoring, with a primary focus on 3D motion estimation for natural hazards. His key research areas include scene flow estimation, computer vision for geoscience, and autonomous sensing of dynamic natural phenomena. Zhu’s major contribution is the development of DeFlow, a pioneering self-supervised model designed to estimate the 3D motion of debris flows—a domain previously dominated by applications in autonomous driving and robotics. To support this work, he introduced a newly captured dataset specifically for debris flow motion, filling a critical gap in automated hazard monitoring. While still early in his career, his 2023 paper has already garnered 5 citations, signaling growing interest from both the computer vision and geohazard communities. Zhu’s work stands out for its novel cross-disciplinary approach, bridging cutting-edge AI techniques with pressing environmental challenges. His research not only advances the field of 3D motion estimation but also holds promise for improving early warning systems and risk assessment for landslides and debris flows, making him a notable emerging voice in applied computer vision.