|
Minh Phuoc Vo
Head of Engineering and Machine Learning
|  |
New: I am co-organizing Three questions about virtual try-on @ CVPR 2025 with Aayush Bansal.
I am the Head of Engineering at SpreeAI. Within my 2.5 years tenure, I transformed the SpreeAI from a struggling $12M valuation company with poor product-market-fit to a $1.5B valuation AI e-commerce company with my virtual tryon brain child. Beyond overseeing all R&D, my greatest pride lies in building and leading a world-class team of passionate machine learning researchers and engineers.Previously, I was a Research Scientist at Meta Reality Labs Research, where I tech-led a group of researchers to develop 3D perception and human sensing algorithms for Meta Aria glasses. Before that, I was a Ph.D. student at The Robotics Institute, Carnegie Mellon University where I worked with Prof. Srinivasa Narasimhan and Prof. Yaser Sheikh on novel methods to capture dense and accurate 3D shape of human bodies. I also worked with Prof. Zhaoyang Wang at the Catholic University of America, where I got B.E. degree in Electrical Engineering, on camera calibration, structured light system, and tracking algorithms.
I am also working part-time with a consulting company, LeCON LLC, to help startups and companies to further develop capabilities in large vision language model (VLM), including data curation, model training, and deployment, and various core components for intelligent systems such as mapping, perception, and prediction. Reach out to us if you are interested in collaborating.
Research
I am very interested in various aspects of 3D vision, physics-based vision, and generative models for photorealistic digitial avatar creation and human scene understanding. The goal is to develop holistic and end-to-end machine learning systems that understand and recreate virtual environments that are perceptually indistinguishable from reality.
Jobs opportuninty: I am hiring full time CV&ML&Graphics researchers. I strike to balance between core and applied research with patents, papers, and product as outputs. Send me an email if you are interested in working with me.
Award
-
Jun. 2024 Ego4D and EgoHuman are selected as 2022-2023 distingished egocentric papers at CVPR's EgoVision workshop.
-
Jun. 2022 Ego4D is included in 33 Best Paper Finalist at CVPR.
-
May. 2017 Won the Qualcomm Innovation Fellowship with Aayush Bansal. See here for the press coverage.
-
Jan. 2014 Our "Stereo Reconstruction" paper (with my previous group) has been awarded Outstanding Paper from Measurement Science and Technology.
Patent
-
Mar. 2022 Time-synchronized Distributed Passive Captures .
-
Feb. 2022 Learning a realistic and animatable full body human avatar from monocular video .
-
Jan. 2020 Scene reconstruction from bursts of image data .
Publication
|
Virtual Fitting Room: Generating Arbitrarily Long Videos of Virtual Try-On from a Single Image
|
|
|
Dress&Dance: Dress up and Dance as You Like it
|
|
|
Efficient Human Vision Inspired Action Recognition using Adaptive Spatiotemporal Sampling
|
|
EgoHumans: An Egocentric 3D Multi-Human Benchmark
|
|
Snipper: A Spatiotemporal Transformer for Simultaneous Multi-Person 3D Pose Estimation Tracking and
Forecasting on a Video Snippet
|
|
IDEO: Large Scale Egocentric 3D Object Dataset and Benchmark Challenges
|
|
TAVA: Template-free Animatable Volumetric Actors
|
|
LISA: Learning Implicit Shape and Appearance of Hands
|
|
BANMo: Building Animatable 3D Neural Models from Many Casual Videos
|
|
Ego4D: Around the World in 3,000 Hours of Egocentric Video
|
|
ODAM: Object Detection, Association, and Mapping using Posed RGB Video
|
|
ContactOpt: Optimizing Contact to Improve Grasps
|
|
ANR: Articulated Neural Rendering for Virtual Avatars
|
|
TexMesh: Reconstructing Detailed Human Texture and Geometry from Monocular Video
|
|
Long-term Human Motion Prediction with Scene Context
|
|
4D Visualization of Dynamic Events from Unconstrained Multi-View Videos
|
|
Spatiotemporal Bundle Adjustment for Dynamic 3D Human Reconstruction in the Wild
|
|
Self-supervised Multi-view Person Association and Its Applications
|
|
|
Occlusion-Net: 2D/3D Occluded Keypoint Localization Using Graph Networks
|
|
CarFusion: Combining Part Detection and Point Tracking for Dynamic 3D Reconstruction of
Vehicles
|
|
Texture Illumination Separation for Single-shot Structured Light Reconstruction
|
|
Passive Tomography of Turbulance Strength
|
|
Automated fast initial guess in digital image correlation
|
|
Hyper-accurate flexible calibration technique for fringe-projection-based
three-dimensional imaging
|
|
Three-dimensional phantoms for curvature correction in spatial frequency domain
imaging
|
|
Advanced geometric camera calibration for machine vision
|
|
|
Accuracy enhancement of digital image correlation with B-spline interpolation
|
|
Phase extraction from optical interferograms in presence of intensity nonlinearity and
arbitrary phase shifts
|
|
Flexible calibration technique for fringe-projection-based three-dimensional
imaging
|
Others
|
Exploiting Point Motion, Shape Deformation, and Semantic Priors for Dynamic 3D
Reconstruction in the Wild
|
External Coverage
