Projects

Student Projects
There is a list of open student projects for 2016/17 available from here. Below, we provide some generic examples of both past and current UG or PhD level student projects.

 

Image Retrieval  Zahra Esbati and Antonia Creswell are working on improving algorithms and technologies for image retrieval. Recently, we have made some progress in applying Generative Adversarial Networks to develop good representations for image retrieval [Creswell & Bharath, 2016]. However, the full retrieval work includes descriptors, hashing and retrieval algorithms.

Image Descriptorssiftwords-150x150  We work on efficient image descriptor technology that is optimised for distinctiveness, adaptability or invariance. Recently, we have been replacing shallow-optimized features with deep techniques.

Control of complex systems using visual inputs  Kai Arulkumaran is working on developing reinforcement learning techniques in deep networks.  Kai and Antonia Creswell have also been collaborating on improving Generative Adversarial Networks.

Medical & Healthcare Applications  Shearin Cao worked on 3D and 4D (either 2D+t or “voxel”+time) image registration and segmentation. Zehra Uslu is currently working on retinal-reflect-150x150training deep networks to perform inference (segmentation, measurements) in retinal images. Graduate Lucas Hadjilucas worked on colonoscopic polyp detection and tracking. Lucas now works at Founders Factory.  Graduate Jose Rivera-Rubio worked on assistive applications of computer vision.  Jose now works for a startup building pipelines for analysing visual data.