AnimalWeb
收藏资源简介:
AnimalWeb是一个大规模的层次化动物面部标注数据集,由起源人工智能研究院创建。该数据集包含21,921个在自然环境下的动物面部图像,涵盖334种不同的动物物种,跨越21个不同的动物分类学目。每个面部图像均标注有9个关键面部特征的标志点。数据集的构建经过四个系统的阶段,涉及超过6,000小时的人工标注努力。AnimalWeb旨在推动动物面部行为自动理解的研究,特别是在动物健康护理和心理学理解方面。
AnimalWeb is a large-scale hierarchical animal facial annotation dataset developed by the Origin Artificial Intelligence Research Institute. This dataset contains 21,921 animal facial images captured in natural environments, covering 334 distinct animal species and spanning 21 distinct animal taxonomic orders. Each facial image is annotated with 9 key facial feature landmark points. The construction of the dataset underwent four systematic stages, involving over 6,000 hours of manual annotation efforts. AnimalWeb aims to advance research on the automatic understanding of animal facial behaviors, particularly in the fields of animal healthcare and psychological understanding.




