Automatic event recognition

Automatic Event Recognition has attracted huge research activity in the last years. It has important applications in both the surveillance industry and search engines for video databases. It comprises of several research topics, including object recognition, object tracking, scene modeling, trajectory modeling and description, and event recognition. In our work, we are designing a novel framework to face the existing challenges. We are using multi-modal techniques to achieve State-of-the-Art object recognition and tracking. Novel techniques are applied to face challenges like object occlusion, changes in lightning, non-rigid object tracking, object distortion due to camera specifications, and multi-camera environments. Moreover, we are introducing the use of semantically rich information about the examined scenes and the object behavior to alleviate the existing challenges.

Demonstration

Relevant Projects

ADVISE
Advanced Video Surveillance archives search Engine for security applications