带标注的疟疾细胞图片数据集

内容 图片采用.png或.jpg格式。一共有3组图像，由1364张图像（约80,000个细胞）组成，不同的研究人员分别准备了这些图像：来自巴西（Stefanie Lopes），来自东南亚（Benoit Malleret）和时程（Gabriel Rangel）。血涂片用吉姆萨试剂染色。 标签 数据包括两类未感染的细胞（RBC和白细胞）和四类感染的细胞（配子细胞，环，滋养体和裂殖体）。注释者被允许将某些单元格标记为困难，即使在单元格类别之一中不清楚也是如此。与未感染的白细胞和感染的细胞相比，未感染的RBC与严重的失衡，占所有细胞的95％以上。 给每个单元格一个类标签和一组边界框坐标。对于所有数据集，Heitor Vieira Dourado热带医学基金会医院疟疾研究人员斯特凡妮·洛佩斯（Stefanie Lopes）给感染细胞赋予了类别标签，表明其发育阶段或被标记为困难。

Dataset Content: All images are in .png or .jpg format. The dataset comprises three groups of images, totaling 1,364 images with approximately 80,000 cells, prepared by three independent researchers: Stefanie Lopes (affiliated with Brazil), Benoit Malleret (affiliated with Southeast Asia), and Gabriel Rangel (for time-course specimens). All blood smears were stained with Giemsa stain.

Annotation Labels: The dataset contains two categories of uninfected cells: red blood cells (RBCs) and leukocytes, as well as four categories of malaria-infected cells: gametocytes, rings, trophozoites, and schizonts. Annotators were permitted to mark certain cells as "difficult to classify", even if their category was ambiguous. Uninfected RBCs exhibited severe class imbalance, accounting for over 95% of all cells when compared to uninfected leukocytes and infected cells.

Each individual cell is assigned a class label along with a set of bounding box coordinates. For the entire dataset, Stefanie Lopes, a malaria researcher at the Hospital Heitor Vieira Dourado Tropical Medicine Foundation, assigned class labels to infected cells to denote their developmental stage, or tagged them as difficult samples.

搜集汇总

背景与挑战

背景概述

该数据集是一个医学图像分类数据集，包含1364张疟疾细胞图片（约80,000个细胞），图片格式为.png或.jpg，由三组不同研究人员采集。数据集提供详细的标注，包括两类未感染细胞和四类感染细胞，每个细胞都有类别标签和边界框坐标，但存在严重的类别不平衡（未感染红细胞占95%以上），并允许标记困难样本。