Quaero

Face Recognition for Smart Interactions

Hazim Ekenel, Karlsruhe University
Face recognition is one of the most important problems of computer vision and pattern recognition. The main group of applications that has fueled intense efforts on face recognition research is security applications, ranging from authentication tasks, such as for access control that can be used in electronic transactions, desktop login, and internet access, to surveillance tasks, such as bank/store and public area security.

In addition to security applications, person identification is one of the most crucial building blocks for smart interaction applications. Here, smart interaction refers to using perceptual technologies for improved human-human and human-machine interactions. Either as an assistant in human-human interactions, e.g. a memory aid that tells the person who he is talking to, or in human-machine interactions, e.g. a machine that recognizes its user and customizes the preferences accordingly, person identification provides the most important characteristic of natural interactions: personalization. Besides, the identity of a person can be used to improve the performances of other perceptual technologies, such as expression analysis systems or head pose estimation systems, by enabling the use of person-specific models.

Face recognition and speaker identification are known to be the most natural person identification methods, since the face and the voice are the modalities that we use to identify people in our daily lives. Although other methods, such as fingerprint identification, can provide better performance, they are not appropriate for natural interactions due to their intrusive nature. The most important advantage of face recognition is the passive identification that it can provide, that is, the person to be identified does not need to cooperate or take any specific action. For example, a smart store can recognize its regular customers while they are entering the store. The customers do not need to talk or look directly into the camera to be recognized. This makes face recognition technology a perfect match for natural interaction applications, since it can work unobtrusively in the background without disturbing or interrupting the subjects to be identified.

Face recognition has found a wide range of smart interaction applications. The application areas, which have been focused on in “Computer Vision for Human-Computer Interaction (CVHCI)” research group at Universität Karlsruhe, can be classified into three groups. Face recognition for smart environments constitutes the first group. This application group corresponds to the identification tasks at a fixed location, for instance, a smart home that identifies the family members. The second group is face recognition for smart machines. In this application group, a machine identifies the subject that it interacts with, for example, a car that identifies its driver or a robot that recognizes the person it serves. The last application group is face recognition for smart image/video retrieval. In this group, face images are used as cues for person retrieval.

Dr. Hazim Kemal Ekenel is working as a researcher at the Institute for Anthropomatics, Universität Karlsruhe, since 2004. After completing his studies in Electrical and Electronics Engineering at Boðaziçi University, Istanbul, Turkey in 2003, he moved to the Universität Karlsruhe, Germany and worked under the supervision of Prof. Alex Waibel and Prof. Rainer Stiefelhagen. His research focuses on face recognition and content-based image and video retrieval. He participates in the projects Quaero, SFB 588 – Humanoide Robots, and CHIL – Computers in the Human Interaction Loop. In 2008 he received the Research Award of the European Biometrics Forum and the Best Demo Award at the IEEE, International Conference on Automatic Face and Gesture Recognition.