TextCaps
收藏arXiv2025-09-30 收录
下载链接:
https://textvqa.org/textcaps
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含了从开放图像数据库中提取的28,000张图片,每张图片都配有五个人工生成的标题,强调了视觉内容中的文本描述。该数据集的任务是进行图像标注。
This dataset comprises 28,000 images extracted from the Open Image Database. Each image is accompanied by five human-generated captions that emphasize the textual descriptions within the visual content. The core task of this dataset is image annotation.
提供机构:
Open Images



