Rat Swarm Optimizer with Graph Convolutional Neural Network Based Waste Object Detector and Classification

Object Detection (OD) is a task of Computer Vision (CV) that includes locating and detecting intended object inside a video or an image. The objective of this OD is not just categorizing the objects but also to sketch bounding boxes surrounding the objects for precise identification of their location in the given visual. Deep Learning (DL) based OD is a CV model which utilizes Deep Neural Networks (DNNs) for automatically locating and detecting inside any video segment or image. It is one among the major crucial progression in the CV field that enables broad applications comprising robotics, self-driving cars, surveillance systems, etc. In this study, a novel Rat Swarm Optimizer with Graph Convolutional Neural Network based Waste Object Detector and Classification (RSOGCNN-WODC) technique is presented. The focus of the RSOGCNN-WODC technique lies in object detection and classification using optimal DL models. To obtain this, the RSOGCNN-WODC technique comprises residual network (ResNet50) model for feature extraction process. Afterward, the RSO system can be exploited for the optimum parameter chosen by the ResNet50 model. At last, the GCNN model is utilized for the recognition of the identified objects proficiently. The performance of the RSOGCNN-WODC system was simulated on the OD database. The results signify the improved results of the RSOGCNN-WODC technique compared to recent DL approaches.