SAgric-IoT: An IoT-Based Platform and Deep Learning for Greenhouse Monitoring

Abstract

The Internet of Things (IoT) and convolutional neural networks (CNN) integration is a growing topic of interest for researchers as a technology that will contribute to transforming agriculture. IoT will enable farmers to decide and act based on data collected from sensor nodes regarding field conditions and not purely based on experience, thus minimizing the wastage of supplies (seeds, water, pesticide, and fumigants). On the other hand, CNN complements monitoring systems with tasks such as the early detection of crop diseases or predicting the number of consumable resources and supplies (water, fertilizers) needed to increase productivity. This paper proposes SAgric-IoT, a technology platform based on IoT and CNN for precision agriculture, to monitor environmental and physical variables and provide early disease detection while automatically controlling the irrigation and fertilization in greenhouses. The results show SAgric-IoT is a reliable IoT platform with a low packet loss level that considerably reduces energy consumption and has a disease identification detection accuracy and classification process of over 90%.

Keywords

• convolutional neural networks
• plant disease
• precision agriculture
• precision farming
• internet of things
• wireless sensor networks