Scholarlink

The increased adoption of face masks in most public and professional spaces has presented wide-ranging challenges to computer vision systems by obscuring critical facial regions. Face mask removal, or face unmasking, has thus become a research trend of revealing the covered facial regions in regard to visual fidelity reconstruction alongside retaining identity. This paper discusses the progress made in this field, more precisely on those which use generative adversarial networks (GANs). We also enumerate datasets which were used or created, and metrics which were used to evaluate the results of different methods introduced in this task. Finally, we state the biggest challenges for research currently while laying out the most promising grounds for future studies.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Face Mask Removal: Datasets, Methods, and Evaluation Metrics

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yazid Safiny,
  • Siham Ahmam,
  • Nidal Lamghari,
  • Abdelghani Ghazdali

摘要

The increased adoption of face masks in most public and professional spaces has presented wide-ranging challenges to computer vision systems by obscuring critical facial regions. Face mask removal, or face unmasking, has thus become a research trend of revealing the covered facial regions in regard to visual fidelity reconstruction alongside retaining identity. This paper discusses the progress made in this field, more precisely on those which use generative adversarial networks (GANs). We also enumerate datasets which were used or created, and metrics which were used to evaluate the results of different methods introduced in this task. Finally, we state the biggest challenges for research currently while laying out the most promising grounds for future studies.