数据堂—399人35,112张黄种人人脸多姿态图像&21关键点标注数据

399人35,112张黄种人人脸多姿态图像&21关键点标注数据集共采集399人，每人88张照片 。数据多样性包括一人多姿态、不同年龄段、不同光照、不同采集环境。399人35,112张黄种人人脸多姿态图像&21关键点标注数据可用于人脸关键点定位、人脸行为识别等任务。

A multi-pose facial image dataset with 21 key-point annotations, which comprises 35,112 facial images collected from 399 recruited East Asian individuals. A total of 88 images were captured for each participant. The dataset exhibits rich data diversity, covering multiple poses per subject, diverse age ranges, varying lighting conditions, and different collection environments. This dataset can be utilized for tasks including facial key-point localization and facial behavior recognition.

搜集汇总

背景与挑战

背景概述

该数据集包含399名黄种人的35,112张多姿态人脸图像，每张图像标注了21个关键点，适用于人脸关键点定位和行为识别任务。数据多样性丰富，包括不同年龄段、光照条件和采集环境，标注准确率超过97%。