Khiem Vuong

I'm a graduate student (MS in Robotics) at the Robotics Institute, Carnegie Mellon University. I'm advised by Prof. Srinivasa Narasimhan (supported by NSF-CPS) as part of the Imaging Group at CMU.

I received my Bachelor's degree in Computer Science from University of Minnesota where I was part of the Multiple Autonomous Robotic Systems (MARS) Lab under the supervision of Prof. Stergios Roumeliotis and Prof. Hyun Soo Park. I spent an amazing summer at Enfusion Systems as a software engineering intern during my sophomore year.

I'm interested in computer vision, machine learning, and robotics. Most of my research is about 3D perception and visual scene understanding.

Email  /  CV  /  Google Scholar  /  Github  /  LinkedIn

• [Mar. 2021] Our paper Deep Multi-view Depth Estimation with Predicted Uncertainty is accepted to ICRA 2021.
• [Jul. 2020] Our paper Surface Normal Estimation of Tilted Images via Spatial Rectifier is accepted to ECCV 2020 as a spotlight presentation.
• [Jun. 2020] Our paper Deep Depth Estimation from Visual-Inertial SLAM is accepted to IROS 2020.

Estimating dense depth and its uncertainty from a sequence of images by leveraging a dense optical flow network to triangulate initial depth maps and refine them through an iterative refinement module.

(Spotlight Presentation)

Surface normal estimation of tilted images by leveraging a spatial rectifier that is learned to transform the surface normal distribution of a tilted image to the rectified one that matches the gravity-aligned training data distribution.

Dense depth estimation from sparse depth points (computed from a Visual-Inertial SLAM system) and RGB images of indoor scenes.

Template from Jon Barron. Last modified: Nov 8th, 2021.