Major crop pests identification research based on Convolutional Neural Network

Abstract

There are many kinds of crop pests in China and they are prone to disasters. Agricultural pests pose a serious threat to crop growth, so how to effectively identify crop pests is crucial. With the development of computer vision technology and artificial intelligence, the combination of computer vision technology and classification and identification of pests has become a hot and difficult point for experts at home and abroad. In this paper, based on the bag-of-words model and the GoogLeNet model, 2200 pest images collected were used as experimental samples to study the identification of crop pests. The experimental results show that the average classification accuracy of the traditional bag-of-words model is about 56.41%, and the GoogLeNet model recognition accuracy can reach 96.35%. The GoogLeNet model based on transfer learning has higher precision and stronger anti-interference ability than the traditional bag-of-words model.