The Humber Care Tech challenge is becoming a highlight of the year for me. I hope they have one next year. The idea is simple enough: Take a problem faced by healthcare professionals and then turn a bunch of techy folks loose to make things better. This year the theme is patients with Alzheimer and related conditions.
The scope of the problem was nicely illustrated by the range of ideas that have emerged. There are 14 teams here today working away and all of them are working on different aspects of the condition, from creating apps to help carers, to helping with shopping, to building reminiscence systems.
Keith and I are working on a system that tries to track the user performing simple processes. The one we’ve chosen is making a cup of tea. We’ve boiled this down (see what I did there?) to movement between a number of fixed kitchen locations (kettle, sink, cupboard etc) and our application will profile the activity in terms of movement between those locations. Once we have observed sufficient repeats of the activity we can detect changes and also detect when someone is having a bad day.
The Open MV environment is wonderful. I’ve not done much image processing before and this is a perfect way to get started. You write the code in Python and then push it into the camera which then runs it. The library is comprehensive and very easy to use. And it runs really quickly.
I’m trying to keep the algorithm silly simple so that it can be made to run on very constrained devices. Essentially I’m using the camera as a glorified PIR sensor. I’m not very proud of the code that I’ve written, but I’m rather proud of the fact that it seems to work.