Design and Development of a Smart Agriculture (SA) System with Machine learning-Based IoT Architecture

Agriculture has always been a crucial part of the economy, but urbanization has started to reduce our fertile land. This rapid reduction has increased the infestations and diseases, bottlenecking production rates and threatening food safety. However, recent advancements in the IoT have posed many potential solutions for this emerging problem. Incorporating IoT-based infrastructures like a smart-farming system can reduce production costs and increase productivity and overall quality while maintaining environmental balance. This study aims to create a smart agriculture system using IoT that can maintain suitable harvesting conditions and classify diseases through a transfer learning-based approach using Deep Convolutional Neural Network (DCNN). Thereon necessary suggestions are passed to the user through the Android app. The system can successfully detect and suggest diagnosis to six classes, including healthy, narrow brown spot, leaf scald, leaf blast, brown spot, and bacterial leaf blight, with an average accuracy of 96.08%. The precision, recall, specificity, and F1-score were 0.9620, 0.9617, 0.9921, and 0.9616 using the non-normalized augmented dataset. The work is expected to offer a roadmap for the future generation of Smart Agriculture Systems by summarizing the cultivation system's disease detection, prevention, and overall optimization process.