Sample Collection and Annotation Method for Intelligent Irrigation Volume Based on Adaptive PID

Intelligent irrigation is one of the key technologies of smart agriculture, which can ensure the on-demand supply of plant nutrients and avoid fertilizer waste. However, the corresponding models are often poorly trained due to the difficulty of accurately labeling the training data. In this paper, we propose an adaptive PID method for annotating irrigation data. According to the drainage situation of plants, the irrigation amount at each irrigation time point is automatically labeled, and experts are asked to confirm whether the labeling is accurate at regular intervals, which reduces the labor cost of expert annotation. The experimental results of tomato cultivation show that the method for data labeling saves the time of expert labeling, and the quality of labeling data is more reliable, and the performance of the trained model is better than that of the ideal irrigation time evaluated by experts. The results show that the proposed method can provide accurate and reliable data support for the machine learning algorithm to predict the irrigation amount, and significantly reduce the workload of manual annotation.