Android-based Corn Disease Automated Recognition Tool Using Convolutional Neural Network

Abstract

One of the most significant crops in the world today, corn, is under attack from several diseases. Typically, visual review and evaluation are used to identify diseases, but they are regarded as being unreliable. Corn farmers needed an automated disease recognition tool to identify different diseases that affect corn. In this study, a pre-trained convolutional neural network (CNN) was employed to create an android-based recognition tool for recognizing corn diseases. A dataset of healthy corn leaves and three (3) maize diseases—common rust, gray leaf spot, and northern leaf blight—was created using an open-source dataset of Plant Village and field images. The researchers used data augmentation, trained the generated neural network, and put it to the test. The dataset was created using a 75–25 split, trained using the transfer learning concept, then fine-tuned using the VGG-16 CNN model. The CNN model was trained using Tensor flow Keras. The model can identify corn diseases, as evidenced by its accuracy of 93.42 percent and F1-score of 93.53 percent. A mobile application employing the Dynamic System Development methodology (DSDM) was created using the methodology. The trained CNN model file was used to create the android application, which serves as a tool for identifying maize diseases. The produced application was deemed to be extremely compliant according to the participants' assessment of the android application using the ISO 25010 software quality standard, with an overall weighted mean of 4.22. The results show that the participants recognized the CDARS application's potential to offer farmers important information and as an automated corn disease recognition tool that could promote more sustainable and secure food production.