MAMe数据集
收藏资源简介:
MAMe数据集是由巴塞罗那超级计算中心创建的一个专注于艺术品媒介识别的高分辨率和可变形状图像数据集。该数据集包含来自三个不同博物馆的数千件艺术品,共计37,407张高分辨率图像,旨在研究高分辨率和可变形状图像属性对图像分类任务的影响。数据集中的分类任务涉及由艺术专家监督的29种媒介(如材料和技术)的区分。MAMe数据集的创建过程涉及从博物馆获取数据,通过专家进行标签映射,并确保数据集的平衡性和高分辨率特性。该数据集的应用领域主要集中在艺术品的自动化分类和识别,以及推动高分辨率和可变形状图像处理技术的发展。
The MAMe dataset is a high-resolution, arbitrarily-shaped image dataset dedicated to artwork medium recognition, developed by the Barcelona Supercomputing Center. It comprises 37,407 high-resolution images of thousands of artworks sourced from three distinct museums, and is designed to study the effects of high-resolution and arbitrarily-shaped image properties on image classification tasks. The classification task included in the MAMe dataset entails differentiating 29 medium categories (e.g., materials and techniques) under the supervision of professional art experts. The creation process of the MAMe dataset involves acquiring data from museums, conducting label mapping with expert guidance, and ensuring the dataset's balance and high-resolution attributes. The primary application scenarios of the MAMe dataset cover automated classification and recognition of artworks, as well as promoting the advancement of high-resolution and arbitrarily-shaped image processing technologies.

- 1The MAMe Dataset: On the relevance of High Resolution and Variable Shape image properties巴塞罗那超级计算中心 · 2021年



