Wei Ye
I am a Staff Research Scientist at Meta Reality Labs, where I have been building the spatial intelligence behind the future of AR/VR since 2020.
Currently, I lead efforts on multimodal LLMs for spatial reasoning, teaching AI to understand and interact with 3D environments. Previously, I led the development of the on-device ML depth estimation system shipped on Quest 3 and Quest 3S. My research lies at the frontier of multimodal large language models, 3D vision, and novel view synthesis.
I received my Ph.D. in ECE from UT Austin, where I was advised by Prof. David Z. Pan, and my B.Eng. from Zhejiang University.
Email / Google Scholar / GitHub / LinkedIn / CV
Publications
See the full list on Google Scholar.
Ri3D: Few-shot Gaussian Splatting with Repair and Inpainting Diffusion Priors
IEEE/CVF International Conference on Computer Vision (ICCV), 2025
CoherentGS: Sparse Novel View Synthesis with Coherent 3D Gaussians
European Conference on Computer Vision (ECCV), 2024
Compressed Depth Map Super-Resolution and Restoration: AIM 2024 Challenge Results
European Conference on Computer Vision Workshops (ECCVW), 2024
Consistent Direct Time-of-Flight Video Depth Super-Resolution
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023
Temporally Consistent Online Depth Estimation in Dynamic Scenes
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023
Multi-Scale High-Resolution Vision Transformer for Semantic Segmentation (HRViT)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022
Earlier Publications
Ph.D. work on machine learning for electronic design automation, 2020 and earlier.
Dealing with Aging and Yield in Scaled Technologies
In Dependable Embedded Systems, Springer, 2020 (Book Chapter)
TEMPO: Fast Mask Topography Effect Modeling with Deep Learning
ACM International Symposium on Physical Design (ISPD), 2020 · Best Paper Award
Re-examining VLSI Manufacturing and Yield through the Lens of Deep Learning
IEEE/ACM International Conference on Computer-Aided Design (ICCAD), 2020
LithoGAN: End-to-End Lithography Modeling with Generative Adversarial Networks
ACM/IEEE Design Automation Conference (DAC), 2019 · Best Paper Nomination
Generative Learning in VLSI Design for Manufacturability: Current Status and Future Directions
Journal of Microelectronic Manufacturing (JOMM), 2019 (Invited)
Litho-GPA: Gaussian Process Assurance for Lithography Hotspot Detection
IEEE/ACM Design, Automation and Test in Europe (DATE), 2019
LithoROC: Lithography Hotspot Detection with Explicit ROC Optimization
IEEE/ACM Asia and South Pacific Design Automation Conference (ASPDAC), 2019 (Invited)
Tackling Signal Electromigration with Learning-Based Detection and Multistage Mitigation
IEEE/ACM Asia and South Pacific Design Automation Conference (ASPDAC), 2019
Machine Learning for Yield Learning and Optimization
IEEE International Test Conference (ITC), 2018 (Invited)
Power Grid Reduction by Sparse Convex Optimization
ACM International Symposium on Physical Design (ISPD), 2018
Placement Mitigation Techniques for Power Grid Electromigration
IEEE International Symposium on Low Power Electronics and Design (ISLPED), 2017
Standard Cell Layout Regularity and Pin Access Optimization Considering Middle-of-Line
ACM Great Lakes Symposium on VLSI (GLSVLSI), 2015
Patents
System and Method for Depth Densification and Confidence Map Generation
U.S. Patent Application 2025/0252531 A1, 2025 (pending)
Holographic Calling for Artificial Reality
U.S. Patent 11,461,962, 2022