Academic Homepage

Computer vision · Robotics · Brain-inspired intelligence

My research develops spatial intelligence for machines, with a focus on 3D vision, robotic perception, SLAM, multimodal sensing, and brain-inspired learning. The long-term goal is to enable machines to perceive, understand, and interact with the 3D world in a human-like way.

Autonomous systems serve as an important research platform: they connect robust perception with spatial reasoning and reliable action in real environments.

Selected publications Prospective students VAST Lab on GitHub ↗

News

Selected updates

  1. ACM TOG 2026: “Closed-Form Convolution for Physically-Accurate Defocus in Gaussian Splatting” was published in ACM Transactions on Graphics. Paper

  2. CVPR 2026: “Linguistic Priors for Visual Decoupling: Towards Symmetric Vision-Brain Alignment” was published at CVPR. Paper

  3. Open-source release: We released a low-cost, portable handheld 3D reconstruction system with build documentation and printable CAD models. Project

  4. CVPR 2025: Our multimodal implicit enhancement method for efficient optical flow estimation was published at CVPR. Paper

Research

Human-like spatial cognition
01 · Perception

Robust multimodal vision

Optical flow, scene flow, low-light vision, multispectral sensing, and sensor fusion for challenging real-world conditions.

02 · Representation

3D world modeling

3D reconstruction, neural radiance fields, Gaussian splatting, pose estimation, and geometric scene representations.

03 · Intelligence

Spatial and embodied intelligence

SLAM, localization, navigation, place recognition, spatial reasoning, and autonomous interaction.

04 · Learning

Brain-inspired learning

Vision–brain alignment, brain-machine coupled learning, multimodal cognition, and biologically inspired representations.

Selected publications

Google Scholar ↗
CVPR2026

Linguistic Priors for Visual Decoupling: Towards Symmetric Vision-Brain Alignment

D. Liu, W. Dai, J. Qian, H. Liu, H. Yi, W. Kong

Paper Scholar
TOG2026

Closed-Form Convolution for Physically-Accurate Defocus in Gaussian Splatting

W. Dai, K. Ma, W. Kong

Paper Scholar
CVPR2025

Multi-Modal Synergistic Implicit Image Enhancement for Efficient Optical Flow Estimation

W. Dai, H. Wu, X. Weng, Y. Zheng, Y. Ming, W. Kong

Paper
ICCV2023

Adaptive Positional Encoding for Bundle-Adjusting Neural Radiance Fields

Z. Gao, W. Dai, Y. Zhang

Paper Scholar
TPAMI2023

Brain-Machine Coupled Learning Method for Facial Emotion Recognition

D. Liu, W. Dai, H. Zhang, X. Jin, J. Cao, W. Kong

Paper Scholar
RA-L2022

Thermal-Inertial SLAM for the Environments with Challenging Illumination

J. Jiang, X. Chen, W. Dai, Z. Gao, Y. Zhang

Paper Scholar
TPAMI2020

RGB-D SLAM in Dynamic Environments Using Point Correlations

W. Dai, Y. Zhang, P. Li, Z. Fang, S. Scherer

Paper Scholar
Complete publication list 39 entries · search and filter

Loading publication index…

VAST Lab

Vision And Spatial Thinking Lab
VAST Lab logo

VAST Lab is my research group at Hangzhou Dianzi University. We study human-like spatial intelligence for autonomous systems and train graduate and undergraduate researchers through both fundamental research and hands-on system development.

Perceive the world. Understand the world. Interact with the world.

Code and projects on GitHub ↗

Teaching and public engagement

Prospective students and collaborators

I welcome master's and PhD applicants, undergraduate researchers, and short-term visiting students interested in spatial intelligence, 3D vision, robotics, or brain-inspired intelligence. Academic and industry collaborations are also welcome.

Send an email University profile ↗