Together with the MENU Research Group from Aalborg University Copenhagen we have been working on a very interesting project – a framework that completely changes the way dietary data is being gathered. Apart from usual geekery and Open Source inspirations we have also tried to employ several usability hacks and see whether they could be of any use in the food area.

The big framework is divided into three distinct, yet closely coupled cores as we did all that was possible to reuse as much of complex code as possible within the project. One of the cores is the Intelligent Buffet that can be seen in the FoodScape Lab. A simple idea – a buffet that tells you what and how much you eat – but if you look into details, a pretty complex setup utilising contactless technology, open source electronics and of course a range of programming languages (C/C++, Python, PHP & bash for scripting).

A normal buffet one could say...

A normal buffet one could say…

The whole setup looks just like a regular buffet that we know from our university or workplace canteens. But don’t be mislead – the inside of our buffet is beefed up with electronics and microcontrollers doing all the magic.

As it was already mentioned – the idea is quite simple – people come in, register, take food and start eating. Apart from the requirement that they have to swipe the chip before they take the food – there is nothing unusual, therefore – minimum bias.

The day the whole setup was tested for the first time we were extremely happy to see that people in fact had no clue that we were knowing that:

  • we know what they take
  • how much they take (+/- 2g)
  • how many times they dig with the spoon into the food dispenser

That is the outcome we were counting on. During the “big testing day” one group of IFS students was trying to answer the question: are people capable of judging how many vegetables they are placing on their plate? A typical guess would be “no”, but as it is in science it is better to be “sure than sorry”. Bear with us – the real answer – backed up with data is coming soon.

Can you tell how many (in grams) vegetables will end up on your plate?

Can you tell how many (in grams) vegetables will end up on your plate?

The buffet was stormed by more than 60 participants – free lasagna and vegetables definitely did the trick. From our perspective the most important was the fact that we had absolutely 0 problems – everything went just smoothly. Obviously that was just a taste of things to come – Murphy’s law was simply delayed for the demo.

As it turned out the real challenge was to interpret properly the overwhelming amount of data and to match everything, so that we could say “person X ate Y of lasagna and Z of vegetables”. A few nights of programming resulted in a few algorithms that can do all this extraction automatically.

Now another interesting bit – the system is controlled via a web page – no need to install any exotic software. A laptop, tablet or anything else with a working web browser can be used to create an experiment and download the data (of course, there is also visualization included).

I ate 380g of lasagna and 44g of vegetables

I ate 380g of lasagna and 44g of vegetables

The buffet is just one part of the big suite – and we have already started working on remaining two which deal with waste management in large volume canteens and that fix data contamination in large dietary studies.

You are probably still waiting for the results of the experiment? Well, people are absolutely not capable of correctly estimating the amount of vegetables they serve themselves on a plate. That means that our awareness of the actual amounts of food we are eating is very low – and that could be a problem when we are presented tasty dishes…