Wednesday, 21 Feb 2018
You are here:

Subscribe to our Newsletter

How many eyes has a typical person? (ex: 1)

Twitter Feed

EMail Print


Claudia Araimo (Eurecom) visiting EPFL
16.04.2012 - 29.04.2012
Internal Visits


With the recent wide adoption of digital video surveillance systems, especially in public spaces and communities, the concern for privacy issues related to such surveillance has increased.  Since video surveillance systems are invasive by their design, it is a hard challenge to find a correct balance between the privacy of the surveyed public and the functionality of the system that helps in the detection of objects and events which give rise to suspicion.  Many different privacy protection techniques have been developed recently, which help to preserve privacy without obstructing the main surveillance objective.  Typically, such techniques assume a reversible distortion, for instance scrambling, of certain regions in video frames that would prevent identification of the objects.  During this exchange, a key visual characteristic of the video that is important to carry the surveillance tasks, but also may contain sensitive information, has been identified.  Furthermore, different privacy protection filters (pixelisation, blurring, masking and scrambling) have been applied.  Finally, objective studies using face detection and recognition algorithms, which are run on the video distorted, with various degrees, by the privacy protection techniques have been conducted.  This research exchange has contributed to WP2, WP5 and WP6 activities.  A poster describing the above, was also presented in the framework of WIAMIS 2012 and MediaSense 2012 in Dublin, Ireland.