AP-10K 动物姿态估计数据集

AP-10K 是首个用于一般动物姿态估计的大规模基准，可用于动物姿态估计的研究。该数据集包括包含大约 1 万张标记有姿态信息的哺乳动物图片，这些图片被归纳为 23 科 和 60 个物种。该数据集遵循分类学等级，带有高质量的关键点标注，并进行了人工检查。

AP-10K is the first large-scale benchmark for general animal pose estimation, tailored to support research in this field. It contains approximately 10,000 mammalian images annotated with pose information, which are categorized into 23 families and 60 species. This dataset adheres to taxonomic hierarchies, features high-quality keypoint annotations, and has undergone manual verification.

搜集汇总

背景与挑战

背景概述

AP-10K是首个用于一般动物姿态估计的大规模基准数据集，包含约1万张哺乳动物图片，涵盖23科和60个物种，并提供高质量的关键点标注。该数据集遵循分类学等级，经过人工检查，适用于相关研究。