Bio-inspired visual attention and object recognition

摘要: This paper describes a bio-inspired Visual Attention and Object Recognition System (VARS) that can (1) learn representations of objects are invariant to scale, position orientation; (2) recognize locate these objects in static video imagery. The system uses modularized algorithms/techniques be applied towards finding salient in scene, recognizing those objects, prompting the user for additional information to facilitate interactive learning. These algorithms based on models human visual attention, search, recognition and learning. implementation is highly modular, modules used as complete or independently. The underlying technologies were carefully researched order ensure they robust, fast, could integrated into an system. We evaluated our system's capabilities Caltech-101 COIL-100 datasets, which are commonly machine vision, well simulated scenes. Preliminary results quite promising our system able process datasets with good accuracy low computational times.