Constructing the Comprehensive Intelligent Monitoring System Based on Wireless Sensor Networks (WSNs) for Mango Crop Precision Agriculture

Abstract

Precision agriculture (PA) is an alternative for increasing agricultural productivity in an efficient and integrated manner. In this study, a comprehensive intelligent monitoring system was developed to capture atmospheric conditions, soil parameters and crop physiology for mango crops, integrating physical sensing devices, object detection models, WSNs and IoT. A physical sensing device validation was conducted to evaluate the performance of the humidity, temperature, soil moisture and soil pH sensors using MAPE analysis, yielding a MAPE of 0.63%. A proposed improved YOLOv10 has been developed to capture the mango fruit condition, whether it has a disease or not, consisting of five classes, including alternaria, anthracnose, black_mould_rot, stem_end_rot and healthy. The improved YOLOv10 leverages the advantages of the StarNet architecture, which is embedded into the backbone structure to achieve high-dimensional features in low-dimensional space, which increases the accuracy of performance without adding computational cost. The improved YOLOv10 achieved an mAP₅₀ value of 0.939, whereas faster R-CNN, SSD, YOLOv5, YOLOv7, YOLOv8, YOLOv10 and YOLOv11 achieved mAP₅₀ values of 0.714, 0.862, 0.880, 0.762, 0.893, 0.905 and 0.921, respectively. Meanwhile, the constructed WSN system implements the Aquila optimiser (AO) method to achieve energy efficiency and an optimal response time, which indicates that at 10 nodes, the energy consumption, delay and loss rate are 165 mJ, 128 ms and 7.6%, respectively. The results of the experiment show that the constructed WSN system has achieved optimal efficiency compared to the existing system.