The effect of lacto‐fermentation over time on the changes in zucchini flesh quality assessed using machine learning models based on image textures

Abstract

AbstractLacto‐fermented foods may have health benefits. Therefore, the consumption of lacto‐fermented fruit and vegetables is important. The objective of this study was to assess the effect of lacto‐fermentation on zucchini flesh after 1 week and 3 months. The zucchini slices were classified using models built based on image textures using different machine learning algorithms. The developed models were characterized by overall accuracies of up to 99.33% for a combined set of textures selected from images in color channels R, G, B, L, a, b, X, Y, and Z (IBk algorithm). Only fresh zucchini slices were correctly classified with an accuracy of 100% and Precision, F‐Measure, Matthews Correlation Coefficient (MCC), Receiver Operating Characteristic (ROC) Area, and Precision‐Recall (PRC) Area of 1.000 for selected models. The greater changes in the zucchini flesh structure were observed after the first week of lacto‐fermentation compared to the fresh slices than between samples lacto‐fermented for 1 week and 3 months.Practical ApplicationsThe obtained results confirmed that the developed procedure combining image analysis and machine learning can be used for the quality assessment of the lacto‐fermented zucchini and the detection of the optimal period to obtain desired changes. The determination of changes in product quality as a result of lengthening the lacto‐fermentation time can be useful for detecting the optimal period of the process.