Plant Growth Prediction through Intelligent Embedded Sensing

Precision agriculture is a research domain aimed at securing the food production for the growing Earth population and urbanization. In this paper, we report on a monitoring system that provides robust prediction of plant growth dynamics. Based on the knowledge about plants growth, we propose a solution that consists of two data analysis stages. The first consists of image preprocessing using filters and time series pruning applying statistical methods. On the second stage we make predictions with superficial machine learning algorithms. The proposed solution can run on embedded devices locally. The computation on low-power devices allows us to reduce the data transmission and ensure `distributed intelligence'. As for the dataset, we collected the top-down view sequences of plant images. We achieved 5% relative error on four days predictions. The experimental results and modelling demonstrate the high potential of the proposed solution for the precision agriculture industry, being a crucial part for the development of a plant growth control system.