ReadingTimeMachine/visual_qa_histograms

这是一个基于直方图的视觉问答数据集，包含合成的单面板直方图图像、用于创建直方图的数据、边界框数据、数据标记以及VQA问答对。数据集分为三个子集，分别包含80、500和1000个图像，每个子集的图像复杂度不同。数据集同时包含了人类和大型语言/多模态模型的注释。

This is a histogram-based visual question and answer dataset containing synthetically generated single-panel histogram images, data used to create histograms, bounding box data, data marks, and VQA question-answer pairs. The dataset is divided into three subsets with 80, 500, and 1000 images each, varying in complexity. The dataset includes annotations from both humans and large language/multimodal models.