RoseVision: An Android Application To Detect Rose Leaf Diseases Using Modified Convolutional Neural Network

Every year 28% of the global cut rose production is hampered by various rose plant diseases. Early detection and classification could diminish the cost of cultivation lost. However, without proper knowledge and expert assistance, farmers are facing challenges to identify the diseases in time to take action. Previously, a few works have been conducted on this. So far, no study has been done to develop an efficient automatic mobile application detection system with faster, lighter, and more accurate predictions. In this study, we presented an artificially intelligent mobile application using image processing to identify rose leaf diseases. The primary goal of this work is to create a simple but exact identification with minimal computing power and time. We proposed a Modified Convolutional Neural Network (MCNN) to classify different rose leaf diseases. The proposed MCNN model outperforms the famous pre-trained models including VGG16, VGG19, DenseNet121, MobileNetV2, Resnet50, InceptionV3, and Extreme Inception (Xception) by achieving 97.94% test accuracy. Additionally, based on the trained MCNN model, we developed an android-based mobile application that will allow the farmers to quickly identify the disease using their android phones.