Deep Learning Based Pest Identification on Mobile

Abstract

Crops, vegetables, fruit trees, flowers and other cash crops, are often harmed by a variety of harmful organisms, plant pathogens, pests, weeds and pest rats, etc. Plant diseases and insect pests often occur, which are one of the main factors which causes the damage of leaves and crop failure. Therefore, in order to stop the pest, it is extremely important to identify the pests of plants and their characteristics correctly. In this paper, an effective and scalable image recognition algorithm is proposed for disease detection. Meanwhile, MobileNets is employed for developing our method on mobile devices. Finally, a dataset consists of three apple diseases is used to demonstrate the effectiveness of our method. In the experiments, transfer learning is used to train a deep convolutional neural network for identifying two types of pest damage, apple rusts and apple Alternaria leaf spot. Our results show that the MobileNets model offer a fast, affordable, and easy-to-deploy strategy for plant disease detection.