AI Deep CNN Models to Detection and Classification of Plan Disease

Abstract

The enormous danger of plant diseases toward agricultural output and food security has made the growth of precise and computerized disease schemes for detection necessary. In this article, a deep learning grounded plant disease category framework is proposed to accurately identify diseases affecting plants from plant photos using a modified VGG-16 architecture. The success rate of the suggested approach was examined and compared with popular deep learning replicas such as CNN, ResNet50 then MobileNet. The experimental findings revealed that the suggested VGG-16 model outstripped other models then achieved the training accurateness level of 97.84% as well as confirmation accuracy of 98.10%. This model achieved the minimum trained loss (0.0821) and a loss of validation (0.0724) compared to all the studied architectures, demonstrating its better learning and generalisation capabilities. CNN achieved, instead, a validation accuracy.