Classification of Tomatoes Growth Degree from Spectrum Hue Information Measured Applying RBF-Kernel Method

In recent years, efforts have been made in the agricultural field to improve crop quality and streamline agricultural management by collecting and visualizing various data on crop growth. In this study, we focus on the “spectrum of light” as information on sunlight that is essential for crop growth and aim to capture its functions. Specifically, a spectral sensor is used to measure the reflectance of light for each wavelength. Measurement data generally contains variations, making classification difficult using machine learning based on linear separation. In this study, we attempt classification using a nonlinear separation method capable of high-dimensional analysis and report its effectiveness. Crop growth varies depending on location, lushness, etc. Traditionally, this work relied mainly on experience and subjectivity, but by quantitatively evaluating the distribution and changes in the visible light band, farmers can visually grasp the growth status of crops and determine the optimal harvest time.