Five of us, one evening a week.

Leads the project and builds the LSTM that reads CNN feature sequences to classify pain. Also designed and built this site.

Trains the CNN that turns each face crop into a 512-dimension feature vector, fixing a class-imbalance bug along the way.

Built patient detection, mapping the body to 17 keypoints, and is now working on fall detection and carer alerts.

Owns face detection and the preprocessing pipeline that turns a raw frame into an aligned crop for the CNN.

Designed the carer dashboard end to end, from early wireframes through to the working prototype shown here.

One seat is held open for a student joining next intake, most usefully someone on embedded or hardware. Email us.