Insect Classification using Deep Convolutional Neural Networks and Transfer Learning On MobileNetV3 Model
Kanwarpartap Singh Gill, Vatsala Anand, Rupesh Gupta, Vivek Pahwa
Abstract
Insect classification is a foundational tool in insect research, providing a basis for understanding insect diversity, ecology, behaviour, and impacts on ecosystems and human activities. Accurate identification and classification of insect species are crucial for advancing our knowledge about insects and their roles in the natural world, as well as for addressing various challenges related to agriculture, biodiversity conservation, pest management, and public health. For the purpose of tracking and recording insect variety, particularly in the context of attempts to conserve biodiversity, insect categorization is crucial. Insect species must be accurately identified and classified in order to understand their distribution patterns, species richness, endemism, and threats—all of which are essential for setting priorities and putting successful conservation plans into action. The subject of entomology offers a decent amount of inventive potential in the creation of advanced deep learning models and the progress of artificial intelligence (AI). In this study, we point to create a exchange learning-based picture classification model that can distinguish different sorts of insect species. With an accuracy rate of more than 80%, our MobileNetV3 model was demonstrated to have strong classification ability for classifying insects.