Zhenyu Li is a final-year Ph.D. candidate in Computer Science at King Abdullah University of Science and Technology (KAUST), where he is advised by Prof. Peter Wonka, specializing in computer vision, 3D vision, and generative AI, with particular emphasis on monocular depth estimation, high resolution 3D perception, and scene reconstruction.
Biography
Zhenyu Li is a Ph.D. candidate in Computer Science at King Abdullah University of Science and Technology (KAUST), where he conducts research under the supervision of Prof. Peter Wonka. Before joining KAUST, he earned his bachelor’s and master’s degrees in Computer Science from Harbin Institute of Technology, China. His research focuses on depth estimation, 3D reconstruction, and geometric scene understanding, with broader interests in generative models and embodied perception. He has gained industrial research experience at Meta Reality Labs, ByteDance Seed, DiDi Cargo, and SenseTime, working on applications spanning AR/VR body tracking, 3D foundation models, and autonomous driving.
Li’s research advances the recovery of detailed and coherent three-dimensional geometry from visual observations. His contributions include BinsFormer for monocular depth estimation, PatchFusion and the PatchRefiner series for high-resolution depth prediction, and Amodal Depth Anything for estimating geometry beyond visible object surfaces. His publications have appeared in leading venues including CVPR, ICCV, ECCV, ICLR, ICML, SIGGRAPH, and SIGGRAPH Asia. During his internship at Meta Reality Labs, he developed EgoPoseFormer V2 for accurate and efficient egocentric human motion estimation, which was selected as a CVPR 2026 Highlight. At ByteDance Seed, he contributed to Depth Anything 3, an ICLR 2026 Oral paper.
Li received the KAUST CEMSE Dean’s List Award in 2024 and the China National Scholarship in 2022. His competition achievements include first place in the ICCV 2023 Visual Continual Learning Challenge’s Multitask Learning for Robustness Track and second place in the ECCV 2022 Mobile AI & AIM Monocular Depth Estimation Challenge. His broader research goal is to develop visual systems that can recover, understand, and reason about the three-dimensional world.
Expertise and Interests
Li's research aims to build intelligent systems capable of perceiving, reconstructing and reasoning about the three-dimensional world from visual observations. His work combines geometric modeling with modern deep learning approaches, spanning depth estimation, 3D scene reconstruction, visual generative modeling, embodied perception and multimodal reasoning. He is particularly interested in developing next-generation foundation models that unify visual perception and language-driven reasoning for applications in robotics, mixed reality and spatial AI.
Awards and Distinctions
- CEMSE Dean's List Award 2024, Computer, Electrical and Mathematical Sciences and Engineering (CEMSE), King Abdullah University of Science and Technology (KAUST), 2024
Education
- Master of Science (M.S.)
- Computer Science, Harbin Institute of Technology (HIT), China, 2023
- Bachelor of Science (B.S.)
- Computer Science, Harbin Institute of Technology (HIT), China, 2021
