Machine learning meets IoT: improved approaches to agricultural leaf monitoring

Abstract

Modern agriculture is experiencing a technological revolution through the integration of sensors and artificial intelligence. These automate critical agricultural operations, including harvest management, crop health assessment, and yield forecasting. At the forefront of this transformation is the Digital Twin Living Laboratory (DTLL) technology, which creates virtual replicas of farms that use real-time sensor networks to improve agricultural decision-making through predictive analytics. The DTLL framework works by deploying comprehensive sensor networks across agricultural environments to capture detailed field and plant data. Advanced data processing algorithms analyze this information to predict and prevent adverse conditions, with the Agriculture Internet of Things (AIoT) serving as the technological backbone of DTLL systems. IoT-based monitoring networks use various types of sensors and imaging systems to track key agricultural parameters, including leaf coloration, ambient and soil moisture levels, and temperature fluctuations. The research presented in this article focuses specifically on viticultural applications, where the technology addresses the critical challenge of preventing fungal and bacterial infections that threaten grape production. By implementing IoT-based monitoring systems in viticultural environments, the study aims to develop predictive models that can identify conditions favorable to the development of pathogens, enabling proactive intervention strategies to protect crop health and maintain production quality. A comparative study of data from leaf and plant sensors with leaf images for the identification of infections and their early prevention is performed.