Data for Deep Learning for Noninvasive Classification of Clustered Horticultural Crops – A Case for Banana Fruit Tiers

This real data has 194 collected banana tier images. It has two parts – the normal and the abnormal (reject) classes, having 139 and 55 samples, respectively. It can be observed the difficulty of classifying banana tiers based only on human recognition due to their clustered and varying physical structures. There are two excel files attached, the data_partition_normal_bananas and the data_partition_reject_bananas, where one can see the train-test data partitions for machine learning or deep learning application. In each file, there are five train-test sets for stratified sampling cross-validation purpose.本真实数据集共包含194张采集得到的香蕉果穗（banana tier）图像，涵盖正常与异常（拒收）两个类别，样本量分别为139与55。由于香蕉果穗呈簇状聚集且物理结构存在个体差异，仅依靠人工识别完成分类存在较大难度。本次附带两个Excel文件，分别为data_partition_normal_bananas与data_partition_reject_bananas，其中存储了可用于机器学习或深度学习任务的训练-测试数据划分方案。每个文件中均包含5组用于分层抽样交叉验证的训练-测试集。