Developing an Advanced IoT-Enabled Smart Agriculture Management System to Enhance Crop Growth: A Hybrid A-Residual-UNet Based Approach

An increasing number of factors, including dwindling arable land, dwindling natural resources, and more intense and unpredictable weather events, have put food security at the forefront of international agendas. The agricultural industry is responding to these challenges by implementing "smart agriculture" strategies that make use of big data and the IoT to increase operational efficiency and productivity. The three main components are feature selection, model training, and preprocessing. For the purpose of preprocessing, the proposed approach use zero unitarization and methods for normalizing diagnostic characteristics that are based on the Quotient Transform formula. Both IG and principal component analysis (PCA) are utilized in feature selection. IG outperforms PCA in terms of accuracy. To make the model training process more accurate, we employed the Hybrid-A-Residual-UNet. With an accuracy of 97.55%, the suggested method outperforms its competitors' methods, such as HDA and ResUNet.