EARLY DETECTION OF STRAWBERRY DISEASES AND PESTS USING DEEP LEARNING

Abstract

This paper presents a deep learning-based classification model for identifying strawberry leaf pest infection in order to be able to identify and cope with the symptoms of pests and diseases at an early stage. The strawberry leaf image data was acquired during the growth period. Due to the insufficient amount of data provided, leaf image data was added through web crawling using the Python library and open data provided by AI Hub. The added data was converted to the same image size to build a dataset, and training dataset and test dataset were defined using pseudo-labeling for stable learning. We applied RegNet and EfficientNet as CNN (Convolution Neural Network)-based image network models for repetitive learning and derived prediction accuracy through ensemble learning. The proposed model helps to identify and deal with pests on strawberry leaves in the growing season at an early stage. Therefore, it is expected to lead the increase in production of the agricultural industry and strengthen competitiveness.