[Unavailable] Cartoon Pose Estimation

[MSc Level] We will collaborate with award-winning animation studio Cartoon Saloon to investigate animated character body part recognition. This project aims to search through cartoon scenes in order to identify character body parts. This task is more difficult than typical human pose-estimation as cartoon characters are often drawn to defy the laws of physics. Knowledge … Read more

[TAKEN] Virtual Conversational Agent with Emotions

Metahuman characters [FYP] This project is about building an emotional conversational virtual agent in Unreal Engine that can take text as input and produce speech and associated animations on a Metahuman. The main aspect will be learning to work with virtual human animations and programming the to animate in state-machines. The project will involve integrating … Read more

[Unavailable] Enfacement for Deep Fakes

CitizenMetaOne Youtuber live capture setup [FYP or MSc level] Become a virtual youtuber, and create your own live Metahuman system setup, using HTC Vive controllers for tracking the hands and a webcam for capturing the face. This project will use pre-existing animation plugins in Unreal Engine as a starting point (Faceware for the facial capture, … Read more

Development environments for education

We are developing a web‐based system for Python programming (called “Pytch”) that targets beginning programmers (see the screenshot below). Because the system is web based a lot of the work will involve JavaScript, TypeScript and React but it’s not necessary to have experience of these at the outset. We have a number of ideas for … Read more

Computers and Sustainability

Happy to discuss student generated ideas and proposals in the area of computers and sustainability and particularly education. Projects could include: Prior projects that have been supervised include: Interested in both Computational Sustainability and Green Computing

[Taken] Robustness of neural networks

The sudden rise of adversarial examples (i.e., input points intentionally crafted so to trick a model into misprediction) has shown that even state-of-the-art deep learning models can be extremely vulnerable to intelligent attacks. Unfortunately, the fragility of such models makes their deployment in safety-critical real-world applications (e.g., self-driving cars or eHealth) difficult to justify, hence … Read more

Privacy and Trust in Federated learning

In the world of big data and artificial intelligence, federated learning is a very attractive way of enabling data sharing in a privacy preserving manner. Clients in a federated learning network never access data from other clients directly and they model parameters they receive are formed from data aggregates.  However, under certain conditions the models … Read more

[Unavailable] Neural Network Cloth Simulation

[MSc] In this project, you will first generate a database of cloth simulation for a single character performing a large range of motions, from a mocap database. You will use a commercial cloth simulation software to create the deformations (e.g., Houdini, Maya, UE4 cloth etc.). Then, the idea is to create a neural network that … Read more