ABSTRACT
Insect pests and crop diseases are the most major variables affecting agricultural production, and reducing agricultural sustainability. It is inconsistent to position security cameras close to the intended pests, and viewing images from the Internet of Things (IoT) monitoring devices at a fixed place is typically insufficient for pest identification. Because of its distinctive capabilities and adaptability in a given setting, such as agriculture, the IoT is a well-known and sophisticated technology and an analytics system that is implemented in many different industries. In order to manage the efficient crop status and crop output, there is a need for the IoT in agricultural areas to reduce the use of chemical fertilizers and crop protection agents. As a result, in this study paradigm, the IoT devices are designated to be used for data collecting. To find and identify the pests in the photos, this study seeks to create a model for pest identification and classification. When the IoT platform is first developed, IoT devices are used to collect data. Then, using YOLOv3, object detection is carried out from the collected photographs to identify the pests in the images. Convolutional Neural Network (CNN) is used to collect deep features from detected images. Convolution Neural Long Short-Term Memory (CNLSTM) is an improved classifier that receives the classified results as pest details. Finally, the Adaptive Honey Badger Algorithm is used to optimize the classifier’s parameters (AHBA). By quickly gathering data about agriculture, the method demonstrates improved performance and ensures technical indications for population estimation and pest monitoring.
