Datasets Comparison

This article creates fruit (banana and guava) image datasets for non-destructive quality classifications. All images were shot 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 original images were collected. The images were augmented to total 8740 images through data enhancement methods (image flipping horizontally, enhancing the image contrast and brightness, boosting the color of the images, and image rotation at 30 degrees). This dataset allows researchers to study different algorithms of machine learning or deep learning for quality classification of fruits.

Computer Vision, Image Processing, Image Classification

Creative Commons Attribution 4.0 International

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.

Computer Vision, Image Processing, Image Classification

Creative Commons Attribution 4.0 International