# Masked Face Recognition Dataset and Application

**Authors:** Zhongyuan Wang, Guangcheng Wang, Baojin Huang, Zhangyang Xiong, Qi Hong, Hao Wu, Peng Yi, Kui Jiang, Nanxi Wang, Yingjiao Pei, Heling Chen, Yu Miao, Zhibing Huang, Jinbi Liang  
**arXiv:** [2003.09093](https://arxiv.org/abs/2003.09093)  
**Date:** 2020-03-20  
**Summary page:** https://www.summarizepaper.com/en/paper/2003.09093v2/

## Key points

- Widespread use of masks is essential in preventing the spread of COVID-19
- Conventional facial recognition technology is ineffective when individuals are wearing masks
- Existing face recognition technology needs to enhance its performance on masked faces
- Three types of masked face datasets have been proposed: MFDD, RMFRD, and SMFRD
- RMFRD is the largest real-world masked face dataset known to date
- These datasets are freely accessible for industry and academia use
- A multi-granularity masked face recognition model achieves an accuracy rate of 95%
- More information can be found at https://github.com/X-zhangyang/Real-World-Masked-Face-Dataset

## Summary

In response to the COVID-19 pandemic, the widespread use of masks has become essential in preventing the spread of the virus. This poses a challenge for conventional facial recognition technology, rendering it ineffective in various scenarios such as community access control, face access control, facial attendance and facial security checks at train stations. To address this issue, there is an urgent need to enhance the recognition performance of existing face recognition technology on masked faces. Most advanced face recognition approaches rely on deep learning and require a large number of face samples. Unfortunately, there are currently no publicly available datasets specifically designed for masked face recognition. To fill this gap, a research team has proposed three types of masked face datasets: the Masked Face Detection Dataset (MFDD), Real-world Masked Face Recognition Dataset (RMFRD) and Simulated Masked Face Recognition Dataset (SMFRD). Among these datasets, RMFRD stands out as the largest real-world masked face dataset known to date. These datasets have been made freely accessible to both industry and academia and serve as valuable resources for developing various applications related to masked faces. Additionally, the research team has developed a multi-granularity masked face recognition model that achieves an impressive accuracy rate of 95%, surpassing industry-reported results. For those interested in accessing these datasets or learning more about their work can be found at https://github.com/X-zhangyang/Real-World-Masked-Face-Dataset The authors of this study include Zhongyuan Wang, Guangcheng Wang, Baojin Huang, Zhangyang Xiong, Qi Hong ,Hao Wu ,Peng Yi ,Kui Jiang ,Nanxi Wang ,Yingjiao Pei ,Heling Chen ,Yu Miao ,Zhibing Huang and Jinbi Liang.