Deep Learning Algorithms for Crop Analysis by Agricultural Experts to Enhance Crop Management and Health
Authors:- Dr.C.Saravanabhavan, Janesh M, Kathirvel T, Dhivagar R
Abstract- – In India, agriculture is an essential industry that makes a substantial economic contribution. However, most of conventional crop monitoring techniques are still done by hand, which makes the procedure time-consuming and ineffective. On the other hand, wealthy countries adopt cutting-edge technologies to increase the productivity of crops and enhance resource utilization. We suggest an integrated strategy for crop health monitoring that makes use of aerial drones, IoT, machine learning, and deep learning in order to close this gap. Several sensory modalities are used in our approach for generating varied information with different accuracy in space, temporal fidelity, and character. While drone-based multispectral imagery collects precise information to create vegetation indices like the Normalized Difference Vegetation Index (NDVI), which calculates crop health based on chlorophyll content, IoT sensors provide real-time environmental data that influences crop development.To obtain a comprehensive analysis, variable-length time-series data from IoT sensors and multispectral images were converted into a fixed-sized representation to generate crop health maps. Several machine learning and deep learning models were applied, with a deep neural network (DNN) with two hidden layers achieving the highest accuracy of 98.4%. Due to the absence of reference data, the health maps were validated through ground surveys and expert evaluations. This technology-driven solution enhances real-time decision-making, optimizing large-scale agriculture in India.
