FREE Virtual Study Fair | 1 - 2 March | REGISTER NOW FREE Virtual Study Fair | 1 - 2 March | REGISTER NOW

Zero-Shot Learning for 3D Point Cloud Segmentation


   Department of Computer Science

  , Prof X Hong  Applications accepted all year round  Self-Funded PhD Students Only

About the Project

Zero-shot learning is the task of learning new classes that are not seen during training. It has received a lot of attention in recent years particularly in deep neural networks (DNNs) based 2D image classification. For 3D data, few researchers have attempted to explore existing 2D approaches to extract meaningful information useful for downstream tasks such as semantic segmentation. However, such direct translation of 2D methods to 3D introduces new challenges and makes the processing of such unstructured 3D data highly challenging particularly due to lack of texture and explicit notion of convolution. In this regards, the objective of this thesis is to first identify the challenges and later develop novel methodologies that makes use of the underlying neighbourhood structure using suitable network architectures (e.g., graph neural networks etc.).

First degree in Computer Science with 2:1 or above MSc degree in the relevant subject areas is desired.


Funding Notes

There is no funding associated with the PhD study. However applicants are encouraged to apply for funding from any funding bodies.

Email Now


PhD saved successfully
View saved PhDs