The Convolutional Neural Network as a Tool for Predicting Fruit Quality and Freshness Based on Images

The lack of technology that can monitor changes in fruit quality in the supply chain can result in a large amount of fruit being wasted. Digital imagery can be used to reduce fruit wastage in monitoring and predicting fruit quality throughout its life. In post-harvest engineering, digital images can be used as virtual representations of real products. This research will present a new approach to monitor the quality changes of banana fruit with machine learning-based imagery, using a thermal camera to acquire data with its ability to detect surface area and physiological changes of banana fruit. In this research, model training has been performed using intelligent technology from SAP after the thermal data dataset has been built. By using thermal information to monitor the status of the fruit, this solution utilizes a deep artificial neural network. The training process has shown that it is more accurate. Therefore, the thermal imaging technique has been used as a data source to create a machine learning-based digital twin of the fruit that can reduce waste in the food supply chain. Thus, 99% prediction accuracy has been achieved.