Medical Plant Identification and Classification Using Average Weight Inertia Based Cat Swarm Optimization and Enhanced Convolutional Neural Network Algorithm

Abstract

Medicinal plants have long been studied and taken into consideration because of how crucial they are to maintaining human health. However, finding medicinal plants takes time, is laborious, and needs knowledgeable professionals. Therefore, the vision-based method may aid both scientists and common people in precisely and swiftly identifying herbaceous plants. In this paper, an upgraded convolutional neural network (ACSO+ECNN) and an average weight inertial cat colony optimisation (ACSO) method are suggested to boost classification and recognition skills. health tree. This work’s four stages include pre-processes, feature extractions and selections with classifications. An adaptive median filter (AMF) is applied during preprocessing to efficiently manage noise. Following that, texture and colour characteristics are extracted as features. Four-dimensional histograms (QH) are used to extract colour data, while grey level co-occurrence matrices (GLCM) are used to extract texture information. Following that, the ACSO method is used to identify significant and pertinent characteristics from the provided image collection. The ACSO algorithm's best likelihood values are used for this. Finally, the ECNN algorithm is used to conduct recognition and classification. The predictions of the ACSO+ECNN model are combined using ECNN to create a model that accurately identifies the proper manufacturer based on a certain kind and set of features. The findings of the experiment suggest the proposed ACSO+ECNN method provides higher accuracy, precision, recall, and minimum execution time than existing methods.