Representing and understanding 3D Objects

Understanding the 3D world around us is a key goal in computer science. If we are endowed with 3D understanding, we can solve problems in several areas in computer science such as robotics, fast 3D modelling and physical reasoning. However, it is not clear how 3D objects can be represented because of the high dimensionality as well as different sources of getting 3D information. Furthermore, in a commercial setting, there are many stages of the graphics pipeline which are very tedious and automating certain parts can significantly simplify the process. The goal of the research is to investigate different methods to represent the shape and structure of 3D objects using deep learning techniques. This knowledge can then be used to create algorithms which can easily do 3D object classification, segmentation and generation. This research will hopefully lead us to have a better understanding of 3D representation which has countless research and commercial applications.

Faculty Supervisor:

David Duvenaud

Student:

Aditya Sanghi

Partner:

Autodesk Canada Co

Discipline:

Computer science

Sector:

Information and communications technologies

University:

Program: