Visual saliency based approach to object detection in computer vision systems: Real life applications

A hybrid approach combining object detection in image and eye-fixation probability map calculation is proposed. This approach has been evaluated on human-like eye-fixation prediction benchmark. Experimental results show viability and efficiency of this approach providing comparison with state-of-art algorithms. The proposed approach has been applied to a real-world application of detection of fire-and-smoke in natural outdoor environment.