3D computer vision

Monocular depth estimation aims to infer the distance information of objects in a 2D image. It is an integral part of many computer vision tasks and has applications to autonomous driving, robotics, and virtual reality, among others. This project focuses on developing a new deep-learning-based monocular depth estimation method with high efficiency, competitive performance, and low generalization error. To this end, various approaches will be explored, including the exploitation of 3D prior and geometric information as well as the design of new loss functions. The project will lead to publications in top computer vision conferences /journals, new datasets, and patents.

Faculty Supervisor:

Jun Chen

Student:

Partner:

New York University

Discipline:

Engineering

Sector:

Education

University:

McMaster University

Program: