BriefGPT.xyz
Apr, 2023
理解偏见需真正均衡的数据集?影响准确性的因素并非身份和图像数量
A Real Balanced Dataset For Understanding Bias? Factors That Impact Accuracy, Not Numbers of Identities and Images
HTML
PDF
Haiyu Wu, Kevin W. Bowyer
TL;DR
本文讨论人脸识别准确率差异的问题,指出虽然许多数据集都平衡了各个人群中身份的数量和图像的数量,但身份和图像数量并不是影响一对一人脸匹配准确性的决定因素,因此需要一个更具有偏差意识的工具包来创建跨人群的平衡评估数据集。
Abstract
The issue of disparities in
face recognition
accuracy across
demographic groups
has attracted increasing attention in recent years. Various face image datasets have been proposed as 'fair' or 'balanced' to assess
→