Development of an algorithm for assessing the antioxidant properties of cherry extracts using artificial intelligence methods

Relevance . Artificial intelligence tools are playing an increasingly important role in food technology and biotechnology, significantly accelerating and improving various processes. Operational control of colorimetric parameters of cherry extracts using computer vision allows for a quick assessment of the content of bioactive substances without the cost of experimental studies. Methods . The objects of the study are samples of cherry extracts. Dilutions from 2 to 0.25% were prepared to obtain images of the test samples. The color characteristics of the extracts were determined using an NR60CP colorimeter. Quantitative determination of the total content of polyphenols in solutions of extracts of different concentrations was carried out using the Folin — Ciocalteu method. RGB images from a digital camera (50 MP) were obtained as input data for classifying objects and compiling a database. The Python programming language, OpenCV library and TensorFlow were used to develop image processing software. TensorFlow extracts features from photographs and adds them to the database. Results . The research results showed that with an increase in the content of bioactive substances — polyphenols — the color of the extracts changes and becomes darker, which is confirmed by the results of the evaluation of color characteristics with a colorimeter: with an increase in concentration, the redness increases and the lightness of the samples decreases. A database was created by extracting features using the TensorFlow library from ready-made sample images. A program for assessing the content of polyphenols in extracts was developed in the Python programming language. The described model of the system for monitoring the quality of extracts using computer vision can facilitate the determination of the content of bioactive substances on an industrial scale, which requires significant time and resources.