Image Datasets of Cocoa Beans for Taxonomy Nuances Evaluation

Six image datasets of cut-test-classified cocoa beans were created from the research presented in [FelipeAntunes_Dissertation_PPGMC.pdf]. Each dataset contains 14 classes (namely: Compartmentalized Brown, Compartmentalized White, Compartmentalized Partially Purple, Compartmentalized Purple, Compartmentalized Slatty, Plated Brown, Plated White, Plated Partially Purple, Plated Purple, Plated Slatty, Moldered, Flattened, Brittle and Agglutinated) with 100 images per class, totaling 1400 images per dataset. Those bases were created to evaluate nuances in classification methods. The source dataset from which those images were derived is present in [An_Image_Dataset_of_Cut-Test-Classified_Cocoa_Beans.pdf].

本研究基于文献[FelipeAntunes_Dissertation_PPGMC.pdf]中的研究成果，构建了6套经切分检验分类的可可豆图像数据集。每套数据集包含14个类别，分别为：分隔型棕色（Compartmentalized Brown）、分隔型白色（Compartmentalized White）、分隔型局部紫色（Compartmentalized Partially Purple）、分隔型紫色（Compartmentalized Purple）、分隔型条板状（Compartmentalized Slatty）、片状棕色（Plated Brown）、片状白色（Plated White）、片状局部紫色（Plated Partially Purple）、片状紫色（Plated Purple）、片状条板状（Plated Slatty）、霉变型（Moldered）、扁平型（Flattened）、脆化型（Brittle）以及黏连型（Agglutinated）；每个类别配有100张图像，单套数据集总图像数量共计1400张。构建此类数据集的目的是评估分类方法中的细微差异。本批图像所源自的原始数据集收录于文献[An_Image_Dataset_of_Cut-Test-Classified_Cocoa_Beans.pdf]中。