GLAMI-1M

GLAMI-1M是由捷克共和国的GLAMI.cz和Rossum.ai创建的大型多语言图像-文本分类数据集，包含111万个记录，每个记录代表一个时尚产品，包括图像、名称和描述，描述使用13种语言之一。数据集用于图像-文本分类，具有191个类别，其中75%的训练集和所有测试集图像由人工标注。该数据集旨在解决电子商务平台中产品分类的问题，并适用于多语言图像-文本分类、文本条件图像生成和多语言机器翻译等研究领域。

GLAMI-1M is a large-scale multilingual image-text classification dataset created by GLAMI.cz and Rossum.ai from the Czech Republic. It comprises 1.11 million records, each corresponding to a fashion product with its associated image, product name and product description, where the description is provided in one of 13 languages. This dataset is designed for image-text classification tasks with 191 categories, and 75% of the training set images as well as all test set images are manually annotated. It aims to solve the problem of product categorization on e-commerce platforms, and is suitable for research fields including multilingual image-text classification, text-conditioned image generation, and multilingual machine translation.