Scene Illustration of Terrestrial Animals with Its Monitoring, Tracking and Recognizing Through Deep Learning in Relation with Granular Computing

Abstract

Combining Deep Learning Technique with Granular Computing employs an inductive paradigm for the terrestrial animal’s elucidation. The proposed method frames the object (terrestrial animal) in arbitrary-shaped and sized granules rather than fixed and rectangular shaped, so that object can effectively mine and recognized. The goal is to present a formal model which automatically focus only on representative pixel of each granule rather than converting pixels from entire image through scanning. Thus, this work entails the process of recognizing not only the static animal in the background, but also depicts moving animal in foreground separately.