SEARISE project pursues three major objectives:
- Achieve comparable to human level of performance in identification, rough categorization, fixation and pursuit of salient events in complex videos for surveillance applications.
- Develop the Smart Eyes prototype.
- Prove the Smart Eyes functionality in real-life environment for observation of large public spaces and restricted in-door areas.
To achieve these challenging objectives, a bunch of new computational models in the field of computer vision will be developed. SEARISE focuses on the development of biologically plausible models of vision:
- Rapid eye-movement control for object fixation and pursuit driven by the visual input.
- Rapid feed-forward mechanism for motion-driven visual attention.
- Construction of the integrated “visual perception – action” loop modulated by visual attention.
- Neural architecture for online learning and recognition of objects and motion patterns.
- Construction of dynamic salience maps.
- Mechanism of dynamic visual attention and object association.
- Feed-back mechanisms for enhancement of learned object shapes and motion patterns ultimately leading to their categorization.
- Integration of feed-forward, cross-streams and feed-back projections into a synchronized neural model.
