数据堂—20,000张人脸图像描述数据

20,000张人脸图像描述数据，包含年龄在18岁以下，18-45岁，46-60岁，60岁以上，的多人种；采集场景丰富，包含室内场景与室外场景；图像内容丰富，含有戴口罩、戴眼镜、戴耳机、表情、姿态以及对抗样本。文本描述语言为英语，主要描述了人种、性别、年龄、拍摄角度、光照以及多样性内容等

This dataset contains 20,000 human face image captioning samples. The included subjects span multiple ethnic groups, with age ranges divided into under 18, 18-45, 46-60, and above 60 years old. The data collection scenarios are diverse, covering both indoor and outdoor environments. The image content is rich, including cases such as wearing masks, glasses, headphones, various facial expressions, postures, as well as adversarial examples. The accompanying text descriptions are in English, mainly describing attributes including ethnicity, gender, age, shooting angle, lighting conditions, and diverse content.

搜集汇总

背景与挑战

背景概述

该数据集包含20,000张人脸图像及对应的英文描述，覆盖多民族、不同年龄段（18岁以下至60岁以上）和多样场景（室内外），内容涉及戴口罩、眼镜、耳机、表情、姿势和对抗样本。描述文本主要涵盖种族、性别、年龄、拍摄角度和光线等信息，通常为30-60词、3-5句，图像为.jpg格式，文本为.txt格式，版权归DataTang所有，可用于商业用途。