Design and Development of Automated IoT-Aided Smart Agriculture Management System for Efficient Crop Growth Using Hybrid Convolution (1D–2D)-Based Adaptive Residual Attention

• Published in: Sensing and imaging 2024-10, Vol.25 (1), Article 60
• Main Authors: Sangeetha, Bathini, Pabboju, Suresh
• Format: Article
• Language: English
• Subjects: Agricultural management; Agricultural production; Algorithms; Convolution; Crop diseases; Crop growth; Crop yield; Economic impact; Electrical Engineering; Engineering; Environmental management; Image enhancement; Imaging; Internet of Things; Irrigation; Medical imaging; Microwaves; Parameter modification; Performance prediction; Pests; Plant diseases; Radiology; RF and Optical Engineering; Smart sensors; Soil structure
• ISSN: 1557-2072; 1557-2064; 1557-2072
• DOI: 10.1007/s11220-024-00512-2

In the agricultural sector, plant diseases are responsible for certain economic losses. So, monitoring the plant's health and detecting plant diseases in the early stages are important to avoid the disease's spread in the entire plant. The classical plant disease detection model faces more complications due to noisy information and inaccurate outcomes in the validation while detecting plant diseases. Hence, to overcome several limitations attained in the conventional disease detection techniques, Internet of Things (IoT) sensors-based smart agricultural model is aimed to develop for monitoring crop growth, pests, and diseases. In this research, three different aspects like detecting plant disease; predicting crop yield, and performing smart irrigation in the agricultural field are introduced. The required crop images, plant images, field images, and soil environmental conditions for smart agricultural management are collected from standard resources using IoT devices. In the first phase, the crop yield prediction is executed by providing the crop image and soil environmental condition to the newly designed structure named a Hybrid Convolution (1D–2D)-based Adaptive Residual Attention Network (H-ARAN). In the second phase, the same H-ARAN is considered for the Plant disease detection phase by providing the plant image and soil and environment conditions as input. In the final phase, the field image and soil and environmental conditions are given to the H-ARAN for predicting smart irrigation. The performance of the H-ARAN in three phases is enhanced by optimizing the parameters using the Modified Chef-based Optimization Algorithm (MCOA). The proposed three phase-smart agricultural management models are helpful to the farmer in enhancing farm productivity and income. Finally, the experimental analysis is conducted on the developed model to confirm its effectiveness. The developed MCOA-H-ARAN-based plant disease detection model secured better accuracy as 95.61% and attained good precision as 95.56%. Hence, the developed framework accomplished superior efficiency in detecting the plant disease and also helps the farmer to protect the entire crop from the pest in the initial phase to benefit with good yield.