Human Posture Classification Dataset

The Human Posture Classification Dataset is designed for applications in visual entertainment and robotics, consisting of a collection of indoor-collected images with high resolutions exceeding 3024 x 4032 pixels. This dataset emphasizes bounding box annotations and tagging to identify half-body portraits and classify them into 14 distinct types of poses, such as crossed hands, hands around the head, and one hand on the cheek, among others.

人类姿态分类数据集旨在服务于视觉娱乐和机器人领域应用，该数据集汇集了高分辨率的室内采集图像，分辨率超过3024 x 4032像素。本数据集特别强调边界框标注与标记，用于识别半身肖像并将其归类为14种独特的姿态类型，例如交叉双臂、双手环绕头部、一手搭于脸颊等。

搜集汇总

背景与挑战

背景概述

Human Posture Classification Dataset是一个包含17,000张高分辨率图像的数据集，专为视觉娱乐和机器人应用设计。数据集通过边界框标注和标签对14种不同的人体姿势进行分类，如双手交叉、手放在头部周围等。