An Efficient System for Real-time Mobile Smart Device-based Insect Detection

Abstract

In recent years, the rapid development of many pests and diseases has caused heavy damage to the agricultural production of many countries. However, it is difficult for farmers to accurately identify each type of insect pest, and yet they have used a large number of pesticides indiscriminately, causing serious environmental pollution. Meanwhile, spraying pesticides is very expensive, and thus developing a system to identify crop-damaging pests early will help farmers save a lot of money while also contributing to the development of sustainable agriculture. This paper presents a new efficient deep learning system for real-time insect image recognition on mobile devices. Our system achieved an accuracy of mAP@0.5 with the YOLOv5-S model of 70.5% on the 10 insect dataset and 42.9% on the IP102 large-scale insect dataset. In addition, our system can provide more information to farmers about insects such as biological characteristics, distribution, morphology, and pest control measures. From there, farmers can take appropriate measures to prevent pests and diseases, helping reduce production costs and protecting the environment.