Download PDF
ABSTRACT
As facility agriculture advances towards high precision and energy efficiency, plant supplemental lighting strategies are shifting from static, preset methods to dynamic, perception-driven approaches. Traditional lighting recipes or empirical supplemental lighting methods often result in plant disease issues, energy waste, and photoinhibition. In recent years, hyperspectral imaging technology has emerged as a powerful, non-destructive monitoring tool, capable of capturing subtle real-time changes in plant photosynthetic pigments, water content, nitrogen levels, and early stress responses. When combined with hyperspectral imaging, machine learning enables the extraction of features and the construction of predictive models from vast spectral datasets, serving as a core driver for the early detection of plant diseases and informed decision-making. This paper systematically reviews recent advances in the integration of hyperspectral technology and machine learning for plant supplemental lighting. Furthermore, it emphasizes the critical role of machine learning models in predicting light demand, diagnosing stress, and addressing plant diseases.
Keywords: Plant Disease; Early Stress; Growth Model; Hyperspectral Technology; Machine Learning