creative-graphic-design/CTXFont

CTXFont（上下文字体）是一个用于研究网页设计中字体选择的数据集。它包含了来自awwwards.com的1,065个专业网页设计，标注了4,893个文本元素。每个文本元素都标注了字体属性（包括字体名称、颜色（RGBA）和大小）以及上下文信息（如HTML标签、设计标签和元素位置）。数据集包含492种独特的字体，并提供了通过自编码器学习的40维字体嵌入。网页设计截图的分辨率为768×1366。该数据集旨在支持基于给定网页设计上下文预测字体属性（字体、颜色、大小）的任务，从而实现考虑设计视觉和语义上下文的自动字体选择系统。数据集分为训练集（4,268个示例）和测试集（625个示例）。

CTXFont (Context Font) is a dataset for studying font selection in the context of web design. It contains 1,065 professional web designs from awwwards.com with annotations for 4,893 text elements. Each text element is annotated with font properties including font face, color (RGBA), and size, along with contextual information such as HTML tags, design tags, and element positioning. The dataset includes 492 unique font faces and provides 40-dimensional font face embeddings learned using an autoencoder. Web design screenshots are included at 768×1366 resolution. The dataset was created for the task of predicting font properties (face, color, size) that match a given web design context, enabling automatic font selection systems that consider the visual and semantic context of the design. The dataset is split into training (4,268 examples) and test (625 examples) sets.