马铃薯质量分级数据集

马铃薯数据集的数据源主要通过网络爬虫获取与Fruits-360[25]剥离的马铃薯数据集组成。按照品种划分包含有荷兰薯、中薯8号、陇薯、大西洋薯、坝薯9号、豫马铃薯1号和川芋5号等系列；按照颜色分为红色未洗马铃薯、红色水洗马铃薯、甜马铃薯、白马铃薯。数据按照剔除芽已经分化的马铃薯、场景较为单一的发芽马铃薯以及非马铃薯图像的规则筛选1098张图像数据，最后取得合乎规则的数据共计513张，加上Fruits-360的1803张马铃薯图像，共计2316张图像；用于创建训练和测试数据集。数据集由马铃薯与发芽马铃薯、腐烂马铃薯构成

The data source of this potato dataset is mainly composed of two parts: potato datasets extracted from Fruits-360[25] and those collected via web crawlers. Classified by variety, it includes series such as Dutch potato, Zhongshu No.8, Longshu, Atlantic potato, Bashu No.9, Yuyang No.1, and Chuanyu No.5. Classified by color, it covers red unwashed potatoes, red washed potatoes, sweet potatoes, and white potatoes. A total of 1098 images were screened according to the criteria of excluding potatoes with differentiated buds, sprouted potatoes with relatively monotonous scenes, and non-potato images, and finally 513 qualified images were obtained. Adding the 1803 potato images from Fruits-360, the total number of images reaches 2316, which are used to construct the training and test datasets. The dataset consists of normal potatoes, sprouted potatoes, and rotten potatoes.