Sightation/SightationVQA

SightationVQA数据集是一个包含4.9k图表图像和29,438个视觉问答对的数据集，旨在为盲人和低视力用户提供优化的图表描述。问答对覆盖了图表的主要结构和细节，并通过有视觉能力的用户反馈来提高描述的质量。The SightationVQA dataset consists of 4.9k diagram images and 29,438 visual question-answer pairs designed to provide optimized descriptions for blind and low-vision users. The question-answer pairs cover both the main structure and details of the diagrams, with quality improvements informed by feedback from sighted users.