Project description
Recently, several systems based on CNNs have been proposed to segment medical images. Systems like V-net, U-net and similar, already show impressive performance on some natural scene images, light microscopy images or MRIs. In this project we will look at the techniques of data augmentation and their impact on the system performance. We may also look at the methods of filter visualization to improve understanding of underlying feature space. The project requires access to a PC with a decent GPU.
Assumed knowledge
Basic knowledge of probability, linear algebra and statistics; good programming skills
Industry involvement
Note: You need to register interest in projects from different supervisors (not a number of projects with the one supervisor).
You must also contact each supervisor directly to discuss both the project details and your suitability to undertake the project.