We present StyleBabel, a unique open access dataset of natural language
captions and free-form tags describing the artistic style of over 135K digital
artworks, collected via a novel participatory method from experts studying at
specialist art and design schools. StyleBabel was collected via an iterative
method, inspired by `Grounded Theory': a qualitative approach that enables
annotation while co-evolving a shared language for fine-grained artistic style
attribute description. We demonstrate several downstream tasks for StyleBabel,
adapting the recent ALADIN architecture for fine-grained style similarity, to
train cross-modal embeddings for: 1) free-form tag generation; 2) natural
language description of artistic style; 3) fine-grained text search of style.
To do so, we extend ALADIN with recent advances in Visual Transformer (ViT) and
cross-modal representation learning, achieving a state of the art accuracy in
fine-grained style retrieval.

本研究提出了 StyleBabel 数据集，包括超过 135K 的数字艺术作品的自然语言标题和描述艺术风格的自由表单标签；通过一种基于 Grounded Theory 的迭代方法进行收集，使用 ALADIN 架构 + Visual Transformer (ViT) 和跨模态表示学习进行 fine-grained style 检索，从而实现了自由形式标记的生成、自然语言描述艺术风格以及精细化文本搜索等多样化的样式特征任务。