MedVQA-GI

该数据集名为MedVQA-GI开发数据集，包含了20,000组图像与文本的配对，每组包括一幅胃肠道内窥镜图像以及描述发现或操作的文字提示。此外，该数据集中包含2,000张独特的图像和483个独特的文本查询，每张图像都与多个提示配对。这个数据集被用于训练和验证，其中随机保留了10%的图像用于验证。其规模达到20,000组图像-文本配对，任务专注于医学领域的文本到图像合成。

The dataset named MedVQA-GI Development Dataset contains 20,000 image-text pairs, each consisting of a gastrointestinal endoscopy image and a text prompt describing clinical findings or medical procedures. Additionally, this dataset includes 2,000 unique images and 483 unique text queries, with each image paired with multiple prompts. This dataset is used for training and validation, where 10% of the images are randomly reserved for validation. With a total scale of 20,000 image-text pairs, the task of this dataset focuses on text-to-image synthesis in the medical domain.