| Precise monitoring of plant growth and environmental parameters maximizes agriculture yield. The traditional monitoring methods are time-consuming and error-prone because of the dynamic and unpredictable plant growth nature. This process has become efficient with the help of artificial intelligence which has automated the process of plant growth monitoring and real-time insights. In plant chambers, greenhouses, controlled environment agriculture (CEA) systems, etc., it is crucial to observe plants and provide them with the required parameters according to their stage which requires farming practices, and accurate information about each plant variety. This paper presents a novel solution for automatic plant identification and monitoring their growth stage by using a dual model approach, i.e. YOLOv8s for plant stage detection and YOLOv11cls for plant identification. Smart Plants incubators (SPI) equipped with the imagery system, sensors, and actuators were used for data collection. An imagery dataset of 7,100 images was collected from smart plant incubators, covering five plants i.e. broccoli, spinach, coriander, tomato, and red radish. The YOLOv8s model achieved mAP@0.5 of 0.963, while the YOLOv11-cls model provided 99.2% accuracy for plant classification despite data imbalance. The system also provides recommendations for plant growth parameters by correlating the detected growth stages with environmental parameters. This developed system helps in efficient resource utilization, optimized operation of sensors and actuators, alert generation for undesired conditions, and real-time monitoring of plants for improved productivity. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.