Histo-DD
收藏资源简介:
Histo-DD是一个针对组织病理学图像的数据集蒸馏算法,由澳大利亚健康创新研究所等机构开发。该算法通过整合染色标准化和模型增强技术,将大型数据集压缩成一组合成样本,以提高训练效率和简化下游应用。Histo-DD主要应用于组织病理学图像的分类任务,通过生成合成样本,减少了训练所需的大量补丁,同时保留了区分性信息,显著降低了训练成本。
Histo-DD is a dataset distillation algorithm designed for histopathological images, developed by the Australian Institute of Health Innovation and other institutions. By integrating staining standardization and model augmentation techniques, it compresses large-scale datasets into a set of synthetic samples, enhancing training efficiency and simplifying downstream applications. Primarily utilized for classification tasks on histopathological images, Histo-DD generates synthetic samples to reduce the massive number of image patches required for training while retaining discriminative information, thus significantly lowering training costs.




