BigDocs/BigDocs-Bench

该数据集包含多个子数据集，每个子数据集针对不同的任务进行配置。这些任务包括将图表转换为标题、Markdown 格式、GraphViz 格式、JSON 格式，以及从 GUI 截图进行各种分析和转换，例如边界框标注、摘要、用户意图分析、视觉问答，还有将图像转换为 SVG 格式、截图转换为 HTML 格式、表格转换为 LaTeX 格式、文本转换为 SVG 格式等。每个子数据集都包含训练集、测试集和验证集。

The dataset consists of multiple sub-datasets, each configured for different tasks. These tasks include converting charts to captions, Markdown format, GraphViz format, JSON format, and performing various analyses and transformations on GUI screenshots, such as bounding box annotation, summarization, user intent analysis, visual question answering. It also includes converting images to SVG format, screenshots to HTML format, tables to LaTeX format, and text to SVG format. Each sub-dataset contains training, test, and validation sets.