Kevin Chen
3D Vision & On-Device Multimodal ML Engineer
I build 3D vision and multimodal ML systems from geometry to edge deployment.
With 8+ years in computer vision, I have built real-time SLAM/VIO, 3D reconstruction, learned visual features, and multimodal inference systems in C++ and Python for resource-constrained XR and AI devices.
Interests: 3D Vision · SLAM & VIO · 3D Scene Understanding · On-Device Multimodal ML
Selected systems
Selected publications
All publications-
CVPR 2025ActiveGAMER: Active GAussian Mapping through Efficient Rendering
Co-authored the CVPR 2025 system and built data preprocessing plus repeatable training and evaluation infrastructure for its Gaussian scene-representation pipeline.
-
ICRA 2024Stereo-NEC: Enhancing Stereo Visual-Inertial SLAM Initialization with Normal Epipolar Constraints
Co-authored work on robust stereo visual-inertial initialization using normal epipolar constraints and bias-aware rotation estimation.