Analyzing Eating Behavior

In collaboration with researchers from Wageningen University and Noldus, we are developing techniques to automatically analyze the eating behavior of people. In particular, we try to determine whether or not people like food that they are eating based solely on cues like facial expressions, body pose, and body dynamics.

We study the behavior of eating people in the Restaurant of the Future, both in controlled and in uncontrolled environments. Our focus in the project is on developing trackers for facial feature points, body pose, cutlery, and food items. These trackers are based on our structure preserving object tracker (SPOT). Using the results of these trackers, we will analyze features such as eating speed, bite rate, etc. and correlate these to human ratings of food appreciation.

For more information on the project, please contact Laurens van der Maaten.

People
Lu Zhang
Laurens van der Maaten
Nicole Koenderink (Wageningen University)
Nico van der Aa (Noldus)