不同缺陷不同影响：人脸识别训练数据的受控审计
原标题：Different Defects, Different Effects: A Controlled Audit of Face Recognition Training Data
AI 摘要
该研究对MS1MV3人脸识别训练集进行受控审计，发现其中仍存在五类数据缺陷：重复图像、身份分裂、裁剪错误、多脸歧义和无效内容。作者在IResNet-50与ArcFace的受控训练协议下分别纠正这些缺陷，发现去重带来的整体收益最明显，合并分裂身份和移除无效内容也能提升训练效率，而联合过滤多种缺陷并未带来同等收益。结果表明数据清洗的效果高度依赖缺陷类型，训练数据质量仍是大规模人脸识别中重要但探索不足的因素。
正文节选
Different Defects, Different Effects: A Controlled Audit of Face Recognition Training Data Abstract The development of face recognition has been strongly supported by the availability of large-scale training datasets. Among them, MS-Celeb-1M has played an important role by providing millions of facial images from nearly 100K identities. However, the size and heterogeneity of such datasets increase the likelihood of annotation errors, redundant samples, and other forms of data inconsistency. This