Quality classification and shelf life determination of spinach using deep learning methodology

Abstract

In this study, a novel deep learning methodology for predicting the shelf life of spinach was proposed. The primary objective of this research was to employ a deep learning approach to determine the shelf life of spinach based on its appearance. The spinach samples were carefully stored at two temperatures, 4 degrees C and 10 degrees C, and the appearance of the spinach samples was regularly recorded using imaging techniques, capturing visual data at various wavelengths. Additionally, total bacterial counts, colour properties and sensorial parameters were assessed. Subsequently, a deep learning model was trained using the collected data. The deep learning algorithms achieved excellent accuracy, with all models surpassing 89.4 % accuracy in predicting food categories. Notably, ResNet-101 algorithm outperformed the others, achieving an accuracy of 93.9 %. This study presents an innovative method for determining the shelf life of perishable food, offering potential benefits that could significantly impact industry practices and enhance consumer well-being. The findings of this study may have practical implications for the food industry, allowing for improved inventory management, reduced food waste and better quality control of spinach products.