Fruits (Banana and Guava) datasets for non-destructive quality classifications

This article provides fruit (banana and guava) image dataset for non-destructive quality classifications. The images were captured with a Redmi Note 10-Pro mobile camera in natural sunlight. All the images were captured at different angles and saved in JPG format. A total of 1738 images were collected. The images were classified into three different classes; Class A, Class B, and Defect Quality, according to the maturity stages of fruits. Further, the image datasets were saved into their respective folder. This dataset allows researchers to study different machine learning and deep learning algorithms for the quality classification of fruits. The dataset provides the foundation for the future study of fruit physiological behaviors underlying postharvest enzymatic browning. Because, in the later stage of the ripening the degradation of fruit peel color from yellow to brown is the major issue in the supply chain and the storage industries.

本文提供了用于无损品质分类的水果（香蕉与番石榴）图像数据集。所有图像均采用Redmi Note 10 Pro手机摄像头在自然日光环境下拍摄，且从不同角度采集，以JPG格式存储，共计1738张。依据水果成熟度阶段，将图像划分为三类：A类、B类以及缺陷品质类，并将数据集按类别存入对应文件夹中。本数据集可供研究人员开展各类机器学习与深度学习算法在水果品质分类任务中的相关研究，同时为后续探究采后酶促褐变背后的水果生理行为奠定了基础。这是因为在果实成熟后期，果皮颜色由黄转褐的劣变是供应链与仓储行业面临的主要难题。