IoT-enabled monitoring and comparative analysis of machine learning techniques for detection of plant diseases in marigolds

Ornamental flower cultivation is highly popular and a critical career that creates income for the farmers. Among these marigold flower is a top-selling flowering plant across most of the countries. Further, this plant has many insect-repellent properties. Despite farming these plants economic benefits, multiple risks are associated with cultivating them. Plant disease plays a critical role which reduces the quality and quantity of flower production flowers and contributes primarily to the loss of production or exports. Manual disease detection strategies are expensive and highly prone to error. Thus, multiple early detection of plant diseases by the Internet of Things (IoT) and Artificial Intelligence (AI) technologies are introduced to reduce the burden of the disease. Most of them are complex and cost high. In conjunction with these, a sensor model suitable for collecting data on marigold leaf diseases is presented in this research. A Machine Learning (ML) technique based on the latest Neural Network (NN) called EFFICIENTNET is employed to isolate the diseased plants from the healthier plants. Experiments conducted and comparisons with existing K-Nearest Neighbor (KNN) techniques showed that the EFFICIENTNET-B0 technique had shown remarkable performance with 97.70% accuracy.