Machine Learning-Based Digital Twin for Monitoring Fruit Quality Evolution

A technological gap to monitor fruit quality evolution in the food supply chain is causing a huge waste of fruits. A digital twin is a promising tool to minimize fruit waste by monitoring and predicting the status of fresh produce throughout its life. In post-harvest engineering, the digital twin could be defined as a virtual representation of real produce. The objective of this work is to present a new approach to create a machine learning-based digital twin of banana fruit to monitor its quality changes throughout storage. The thermal camera has been used as a data acquisition tool due to its capability to detect the surface and physiological changes of fruits throughout the storage. In this study, after constructing the dataset of thermal data belonging to four classes, the training of the model has been performed using intelligent technologies from SAP. The solution has applied a deep convolutional neural network to monitor the fruit status based on the thermal information, and the training process has shown higher accuracy. Thus, 99% of prediction accuracy has been achieved which is proved to be a promising technique for the development of fruit digital twins. The application of thermal imaging techniques can be used as a data source to create a machine learning-based digital twin of fruit that can minimize waste in the food supply chain.