WeChat mini program for wheat diseases recognition based on VGG-16 convolutional neural network

Abstract

To solve the problem of pesticide misuse in judging common wheat diseases, we propose a wheat disease identification scheme combining the VGG convolutional neural network model with WeChat mini program technology. In the model training process, we continuously adjust the structure of the convolutional neural network VGG-16 to realize the real-time and accurate identification model of wheat disease. Specifically, the model parameters are optimized by locally adjusting the convolution layer of the VGG-16 network to achieve accurate maximization. Through the verification, the best accuracy rate of the wheat disease identification model is 85.1%. The mini program is compiled by the WeChat developer tool, which is developed based on WXML, WXSS and JavaScript. After building the wheat disease identification model, it is deployed on a cloud server that works continuously for working 24 h a day. In addition, the mini program posts HTTPS requests as the function of wheat disease identification. The implementation of this scheme can help users identify different types of wheat diseases and provide corresponding solutions according to the results, which is of great significance in underdeveloped agricultural areas.