Bootstrapper recognizes tabletop computer users by their shoes

Bootstrapper uses depth cameras to capture images of a user's shoes to compare against a database of known shoe images
Image Gallery (2 images)

Facial recognition might be all the rage in giving computer systems the ability to ascertain the identity of individuals - what with most people having different facial features and all. But a team from the Hasso Plattner Institute in Potsdam, Germany, has taken a different approach to identify users of touch-based tabletop computers like Microsoft's Surface. Instead of focusing on the face, the team has looked in the opposite direction to develop a system known as Bootstrapper which distinguishes between users based on their footwear.
When a user interacts with the tabletop computer, the Bootstrapper system, which consists of one or more depth cameras mounted to the table's edge, observes their shoes and matches them to a database of known shoe images that are associated with specific user profiles. When multiple users are interacting with the table at the same time, the system also takes into account the hand orientation of the touch inputs so they aren't mismatched.
The team, which includes Patrick Baudisch, a professor of computer science, and graduate students Stephan Richter and Christian Holz, has developed a prototype of the Bootstrapper using a Kinect and claim that it can recognize individuals from a database of 18 users with 89 percent accuracy.

Obviously the system has some shortcomings. Two people wearing the same type of shoe or one person wearing different shoes at different times will render the system useless. However, the team says it chose such an approach because shoes offer distinct features - color, texture, design, etc. - and, because shoes are generally aligned with the ground, they are easier to track.
Additionally, the system isn't intended to act as a gatekeeper to secure systems, but rather for things such as keeping track of the progress of students in a classroom environment.