英文摘要：

Design of Agricultural Internet of Things System for Microgreens Growing Environment

Abstract: With the development of urban agriculture and precision farming, microgreens have drawn increasing attention due to their high nutritional value and short growth cycle. To enhance the intelligence level of cultivation processes, this study designs and implements an agricultural Internet of Things (IoT) system that integrates environmental monitoring, image recognition, and remote control, aiming to provide technical support for efficient microgreen production. Focusing on radish microgreens, the study builds an embedded hardware platform based on Raspberry Pi 4B and ESP32, enabling real-time monitoring and precise regulation of key environmental parameters such as temperature, humidity, and light intensity. A high-quality image dataset covering sprouting, seedling, and mature stages is constructed. Lightweight convolutional neural networks including EfficientNetV2, MobileNetV3, and ShuffleNetV2 are employed for growth stage classification, with Grad-CAM used for visualizing model decision-making mechanisms. Furthermore, a cloud platform based on the MQTT protocol is developed to achieve seamless data interaction and remote control between edge nodes and the cloud. Experimental results demonstrate that the proposed system performs well in terms of recognition accuracy, environmental regulation responsiveness, and data transmission stability. The study provides a feasible approach for digital management of micro-scale crops in smart agriculture applications.

Keywords: Agricultural IoT; Deep Learning; Microgreens; Environmental Monitoring; Image Recognition; Grad-CAM

Classification: TP393.08, TP391.41, S436.13