Chinese Face and Body Dataset: lab and personal photos of the same individuals

Collecting personal photos. The guidelines for personal photo collection were asfollows. Participants were required to provide a total of 10 personal photos that meet specificcriteria to ensure diversity and comprehensiveness. These photoswere required to vary in type,capturing different times and angles, such as front and side views. They also had to reflectdifferent facial expressions, including smiling and neutral, and could be taken with and withoutmakeup. Additionally, the photos had to feature various backgrounds, encompassing bothindoor and outdoor settings, and different lighting conditions. Participants were instructed towear different attire in these photos, including both casual clothes and uniforms. The photosneeded to be named according to the participant’s age at the time the photo was taken.Along with the photos, participants were required to fill out a basic demographic informationform. Note that makeup and accessories were allowed in our dataset. Photos that displayedissues—such as obstructions covering facial features, excessive similarity affecting individual distinctiveness, blurriness, or unrecognizable facial features—prompted experimenters torequest additional images. The photos with issues were included in the dataset.distinctiveness, blurriness, or unrecognizable facial features—prompted experimenters toObtaining laboratory photos. In the laboratory photo collection, model participantsvisited the lab to have standardized photos taken. They were instructed not to wear heavymakeup during the photo sessions. The photos were captured using a Canon EOS 6D MarkII digital camera with a 50 mm lens, set to a resolution of 72 dpi and a size of 4160×6240pixels. The camera was positioned at an adjustable height to align with the model’s face,ensuring that individuals of varying heights were fully visible in the photos. A green curtainserved as a consistent background, while two softboxes, placed one meter on either side ofthe model, provided bright and evenly distributed illumination.During the laboratory photo session, model participants were photographed under nine ∗specific conditions. These conditions were categorized across fourvariables: photo types (whole body and headshots), pose types (spontaneous and standardpose), expression types (neutral and smiling), and viewpoints (frontal, 45-degree left profile,45-degree right profile, and back). This categorization resulted in nine distinct image types,such as a whole-body standardized pose with a neutral expression from a frontal viewpoint.

个人照片采集。个人照片采集的规范如下：参与者需总共提供10张符合特定标准的个人照片，以确保样本的多样性与全面性。这些照片需在类型上具备差异，涵盖不同拍摄时段与拍摄角度，例如正面与侧面视角；同时需展现不同的面部表情，包括微笑与自然中性表情，且可化淡妆或素颜拍摄。此外，照片需包含多样的拍摄背景，涵盖室内与室外场景，以及不同的光照条件。参与者需在各照片中身着不同服饰，包括休闲装与制服。照片需按照拍摄时参与者的年龄进行命名。除照片外，参与者还需填写一份基础人口统计学信息表格。请注意，本数据集允许照片中出现妆容与配饰。若照片存在遮挡面部特征、相似度偏高影响个体辨识度、画面模糊或面部特征无法辨识等问题，实验人员将要求参与者补充拍摄照片。存在上述问题的照片仍将被纳入本数据集。

实验室照片采集。在实验室照片采集环节，参与者需前往实验室拍摄标准化照片。拍摄期间，参与者不得浓妆艳抹。本次拍摄采用佳能EOS 6D Mark II（Canon EOS 6D Mark II）数码相机搭配50mm镜头，设置分辨率为72 dpi，图像尺寸为4160×6240像素。相机高度可调节，以匹配参与者的面部高度，确保不同身高的参与者都能完整出现在画面中。拍摄背景为统一的绿色幕布，同时在参与者两侧各1米处布置两盏柔光箱，提供明亮且分布均匀的照明。

在本次实验室拍摄环节中，参与者需在九种特定拍摄条件下完成拍摄。这些条件可分为四大变量：拍摄景别（全身照与头部特写）、姿势类型（自然姿势与标准姿势）、表情类型（中性表情与微笑表情），以及拍摄视角（正面视角、左侧45°侧脸、右侧45°侧脸与背面视角）。上述分类共生成九种不同的图像类型，例如正面视角下的全身标准姿势中性表情照片。

搜集汇总

背景与挑战

背景概述

该数据集包含中国个体的面部和身体照片，分为个人照片和实验室照片两部分，旨在提供多样化和标准化的图像数据。个人照片由参与者提供，覆盖不同时间、角度、表情、化妆、背景和着装，增强了现实场景的多样性；实验室照片在受控环境下拍摄，使用专业设备确保一致性和高质量，涵盖九种特定条件（如照片类型、姿势、表情和视角）。数据集适用于人脸识别、身体分析和相关计算机视觉研究，强调个体在自然与实验室环境下的对比。