Monitoring and tracking human movements and recognizing the type of activity is a current “hot topic” in the field of computer engineering. A specific application of such technology would be in sports science whereby the movements and actions of a person are automatically detected and recognised using a platform as simple as a mobile phone. Essentially monitoring human exercise can be greatly improved with the use of computer vision, by automatically determining how a user is exercising and providing feedback on how to improve the efficacy of that exercise. This involves a field of computer vision focused on activity recognition.
We are therefore interested in exploring ways of achieving real-time activity recognition, with the important pipeline stages of joint detection, pose estimation, and joint tracking. Balancing computational efficiency with accuracy is important, as users may want to be able to execute these algorithms on embedded devices such as smartphones. This is a project with strong commercial applications and an industry sponsor, and further helps develop good academic research skills.
Undergraduate
None
Lab allocations have not been finalised