Integrating IoT and Image Processing for Crop Monitoring: A LoRa-Based Solution for Citrus Pest Detection

Abstract

Today, agriculture faces many challenges, such as the use of inefficient methods that affect crop quality. Precision agriculture (PA), combined with advanced technologies, improves monitoring, while the integration of wireless communication optimizes processes and resources. This work presents the design of a communication prototype applied in precision agriculture, which allows the acquisition, processing, and wireless transmission of information extracted from the Cotonet pest to The Things Network (TTN) cloud server. This prototype integrates technologies and protocols such as LoRaWAN, Message Queuing Telemetry Transport (MQTT), Internet of Things (IoT) sensors, and Computer Vision. This prototype employs a robust processing and segmentation algorithm, which allows the recognition of pests in citrus plants based on color. The results show that lighting conditions, weather, and time of day influence the quality of the captured images. The relationship between image resolution, brightness, and processing time shows that higher-resolution images (1920 × 1080 pixels per image) provide better detection of pest pixels (greater than 50% of the pest index) but require longer processing time (28.415 ms on average). Furthermore, the developed system effectively detects an index of affection of Planococcus citri (Cotonet) in agricultural plantations through an end-to-end technological implementation that integrates image processing, wireless communication, and IoT technologies.