PARE dataset
收藏arXiv2022-10-14 更新2024-06-21 收录
下载链接：
https://github.com/davidcferman/pareidolialandmarks
下载链接
链接失效反馈官方服务：
资源简介：
PARE数据集是一个包含150张图像的小型数据集，专注于pareidolias（错觉脸），这些图像抽象地类似于人类面孔。该数据集用于展示所提出方法在处理新颖且具有挑战性的面部定位问题上的有效性。数据集中的图像来源于现实世界，具有多样性和复杂性，旨在推动面部定位技术的发展，特别是在处理非典型面部形态和极端条件下的应用。
The PARE dataset is a small-scale dataset containing 150 images focused on pareidolias (illusory faces), which abstractly resemble human faces. This dataset is used to demonstrate the effectiveness of the proposed method in addressing novel and challenging facial localization tasks. The images in the dataset are sourced from real-world scenarios, featuring diversity and complexity, aiming to advance the development of facial localization technologies, especially for applications involving atypical facial morphologies and extreme conditions.
提供机构：
AI Foundation, USA 2 UT Austin, USA
创建时间：
2022-03-20
