MacPaw/Screen2AX-Task

Screen2AX-Task数据集是Screen2AX系列数据集的一部分，旨在通过计算机视觉和深度学习提高macOS应用程序的可访问性。该数据集专注于UI任务定位，将macOS应用屏幕截图与任务描述及其对应的视觉参考相结合。每个样本包括一个应用屏幕截图、一个目标UI区域的边界框、原始图像尺寸、一个任务描述和一个视觉描述。数据集适用于语言定位、UI元素关联和视觉语言模型训练等任务。

Screen2AX-Task is part of the Screen2AX dataset suite, which is a research-driven collection aimed at advancing accessibility in macOS applications using computer vision and deep learning. This dataset focuses on UI task grounding, pairing macOS application screenshots with task descriptions and their corresponding visual references. Each sample consists of an application screenshot, a bounding box for the target UI region, dimensions of the original image, a task description, and a visual description of the UI target. The dataset is designed for tasks such as language grounding, UI element linking, and vision-language model training for accessibility applications.