MangoLeafVarietyBD: An Image Dataset for Multi-Class Mango Variety Classification in Bangladesh

The MangoLeafVarietyBD (Raw Images) takes 2,744 original, unprocessed photos of mango leaves directly taken in the field of Thakurgaon District and Daffodil International University (Savar, Dhaka), Bangladesh. The photos were taken on Samsung Galaxy S24 Ultra and Google Pixel 8 and the background was covered with a white piece of paper to make sure that the visuals were consistent across the photos of every leaf.
Eight varieties of mangoes, including Amrapali, Banana, Bandigori, Himsagar, Surjapuri, Harivanga, Kacha Mitha, and Brunei King are included in the dataset. Each of the images has its label as the class name and is then sorted into class-specific folders that can be used in PyTorch or TensorFlow ImageFolder, allowing the reproducible preprocessing and allowing a fair comparison with other mango variety classification systems designed to be used in a field.