WBCAtt

WBCAtt数据集是由南洋理工大学新加坡的研究人员创建的，专注于白血球的详细形态学属性标注。该数据集包含10298张图像，每张图像都标注了11个形态学属性，共计113278个标签。这些属性是通过与病理学家的讨论、文献回顾和手动检查显微镜图像确定的，旨在解决白血球分析中可解释人工智能（XAI）的当前差距。数据集的应用领域包括提高白血球识别模型的解释性，进一步推动病理学和血液学领域的XAI发展。

The WBCAtt dataset was developed by researchers from Nanyang Technological University, Singapore, focusing on detailed morphological attribute annotation of white blood cells. This dataset consists of 10,298 images, each annotated with 11 morphological attributes, totaling 113,278 labels. These attributes were determined through discussions with pathologists, literature reviews, and manual examination of microscopic images, aiming to address the current gaps in explainable artificial intelligence (XAI) for white blood cell analysis. The dataset's application scenarios include enhancing the interpretability of white blood cell recognition models and further advancing the development of XAI in the fields of pathology and hematology.

搜集汇总

背景与挑战

背景概述

WBCAtt数据集是一个专注于白血球形态学属性标注的数据集，由南洋理工大学新加坡的研究人员创建，包含10298张图像，每张图像标注了11个形态学属性，总计113278个标签。该数据集旨在通过详细的属性标注解决白血球分析中可解释人工智能（XAI）的差距，应用于提高白血球识别模型的解释性，推动病理学和血液学领域的XAI发展。