Work

Predicting postharvest weight loss and texture changes in table grapes using fruit color and machine learning

Year: 2025

Type: article

Abstract: Accurately predicting postharvest quality is crucial for optimizing storage and reducing losses in table grapes. This study explores the potential of fruit color parameters as non-invasive indicators of postharvest weight loss and textural changes. Using convolutional neural networks (CNNs), we developed predictive models based on colorimetric data, achieving high accuracy (R² > 0.85 for weight loss and R² > 0.96 for texture). Additionally, the effects of storage temperature on grape quality were examined, revealing that cold storage (3 °C) significantly reduces weight loss and maintains texture better than storage at 10 °C. Among tested cultivars, ‘Shine Muscat’ exhibited superior moisture retention and textural stability compared to ‘Flame Seedless’. These findings highlight the potential of integrating color-based assessments and machine learning models into postharvest monitoring, offering a practical approach for improving quality control and storage management in the grape industry.

Source: Future Foods

DOI: 10.1016/j.fufo.2025.100703