Screen2AX-Tree, Screen2AX-Element, Screen2AX-Group
收藏资源简介:
Screen2AX数据集由乌克兰天主教大学和MacPaw公司的研究人员创建,包含112个macOS应用程序的界面截图及其对应的UI元素检测、分组和层级可访问性元数据。数据集旨在为可访问性生成研究提供基础,帮助开发人员自动创建实时的、树状结构的可访问性元数据,从而提高桌面应用程序的可访问性。
The Screen2AX dataset was created by researchers from the Ukrainian Catholic University and MacPaw Inc. It includes interface screenshots of 112 macOS applications, along with their corresponding UI element detection, grouping, and hierarchical accessibility metadata. This dataset aims to provide a foundation for research on accessibility generation, helping developers automatically generate real-time, tree-structured accessibility metadata to enhance the accessibility of desktop applications.
Screen2AX 数据集概述
数据集简介
Screen2AX 是一个基于计算机视觉和深度学习的项目,旨在生成 macOS 应用程序的无障碍访问功能。相关研究论文可在 arXiv 查阅。
数据集列表
相关模型
系统要求
- macOS
- Python (推荐 ≥ 3.11)
- Conda
- Pip
安装与使用
-
创建并激活项目环境: bash conda create -n screen2ax python=3.11 conda activate screen2ax pip install -r requirements.txt
-
无障碍生成脚本: bash python -m hierarchy_dl.hierarchy --image ./screenshots/spotify.png --save --filename spotify.json
-
屏幕阅读器: bash python -m screen_reader.screen_reader --name Spotify
许可证
- YOLO 模型:GNU Affero General Public License (AGPL)
- BLIP 模型:MIT License
- 数据集:Apache 2.0 license
- 代码库:MIT License
引用
bibtex @misc{muryn2025screen2axvisionbasedapproachautomatic, title={Screen2AX: Vision-Based Approach for Automatic macOS Accessibility Generation}, author={Viktor Muryn and Marta Sumyk and Mariya Hirna and Sofiya Garkot and Maksym Shamrai}, year={2025}, eprint={2507.16704}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2507.16704}, }

- 1Screen2AX: Vision-Based Approach for Automatic macOS Accessibility Generation乌克兰天主教大学,利沃夫,乌克兰;MacPaw,基辅,乌克兰 · 2025年



