An Artificial-Intelligence-Based Novel Rice Grade Model for Severity Estimation of Rice Diseases

The pathogens such as fungi and bacteria can lead to rice diseases that drastically impair crop production. Because the illness is difficult control on a broad scale, field monitoring one of most effective methods control. It allows for early detection disease implementation preventative measures. Disease severity estimation based digital picture analysis, where pictures are obtained from using mobile devices, strategies. This paper offers method quantifying three (brown spot, blast, bacterial blight) determine stage plant disease. A total 1200 images illnesses healthy make up input dataset. With help agricultural experts, diseased zone was labeled according type Make Sense tool. More than 75% in dataset correspond label, plants represent more 15%, multiple 5% labeled. proposes novel artificial intelligence grade model uses an optimized faster-region-based convolutional neural network (FRCNN) approach calculate area leaf instances infected regions. EfficientNet-B0 architecture used backbone shows best accuracy (96.43%). performance compared with CNN architectures: VGG16, ResNet101, MobileNet. evaluation parameters measure positive predictive value, sensitivity, intersection over union. be further deployed tool farmers obtain perfect predictions level lesions conditions produce crops organically.