Outdoor Environment Classification

Abstract

Environment classification is extremely useful for Robotic applications and visually impaired person. The core objective of this exertion is to identify the background environment of the scene based on the occurrence probability of the objects present in the scene. Various approaches have been proposed for Scene classification and recognition; we propose a novel RetinaNet with EnvClaObjectness Algorithm for environment classification, RetinaNet achieves greater object detection accuracy. EnvClaObjectness Algorithm, exploit occurrence probabilities of the recognized objects to classify the environment.