Background
Students at Linköping University bring lunch boxes with them which they want to heat. This is possible to do in student kitchens, which are placed on various locations around campus. Some students feel like they often are too crowded and that it thereby takes too long to get the food heated. Even if a particular kitchen has a queue, many students do not consider going to another kitchen in case it is even more crowded. A practical solution would be a mobile service in which students can se the occupancy, as well as queue severity of the campus kitchens in real time.

Further more, measurements of kitchen occupancy and usage provide empirical data for the school boards when making decisions regarding student kitchens, among other things the construction of new ones.

Concrete problem formulation
What is required is a a way of reliably separating, tracking and counting of people from above, in places where the sensors have to be very close to the persons tracked, thereby  having a limited field of view. Furthermore, the solution must be running in real time and not require to much processing power, since the ultimate goal is to run it on cheap embedded devices in the future.

Solution