In modern advertising, eye-tracking software and equipment can be very useful to determine how many passing pedestrians are looking at your poster or sign. The problem with most of this expensive eye-tracking equipment is that is needs to be calibrated and does not always work in real-world situations, only capturing eye contact if the person looks at a specific area – usually right at the camera itself. The problem gets worse when crowds of people walk past or the light quality becomes poorer as it gets dark.

Researchers from Osaka University, the Excellence Cluster at Saarland University and the Max Planck Institute for Informatics think they have cracked the problem with new software that can reliably detect eye contact from passersby as they glance at an advertisement. This new software estimates gaze direction using a type of neural network known as deep learning.

"Until now, if you were to hang an advertising poster in the pedestrian zone, and wanted to know how many people actually looked at it, you would not have had a chance," says Andreas Bulling, the research group leader.

[ https://www.youtube.com/watch?v=ccrS5XuhQpkOpens in new window ]