2,937人面部遮挡多姿态人脸识别数据【数据堂】

2,937人面部遮挡多姿态人脸识别数据，每位被采集者，分别采集在10种遮挡条件下（包括不遮挡条件）*4种光线下*5种人脸姿态，共计10*4*5=200（张）人脸数据，1,930人面部遮挡多姿态人脸识别数据可应用于遮挡人脸检测及识别等计算机视觉任务

This is a face recognition dataset involving 2,937 individuals with occluded and multi-pose faces. For each subject, face images were captured under 10 occlusion conditions (including the non-occluded condition), 4 lighting conditions, and 5 facial poses, resulting in a total of 10×4×5=200 face images per individual. The 1,930-individual subset of this dataset can be applied to computer vision tasks such as occluded face detection and recognition.

搜集汇总

背景与挑战

背景概述

该数据集包含2,937人的面部图像，每人200张照片，总计约587,400张图像，专门设计用于遮挡人脸检测和识别任务。数据覆盖10种遮挡条件、4种光线和5种姿态，具有高多样性，包括不同年龄、性别和采集环境，且标注准确率超过97%。需要注意的是，这是商业数据，仅限企业合作购买使用。