FaceMask

A face-mask detector dataset for performance evaluation and test benchmarking

The FaceMask Database is publicly available and it can be used for experimental purposes with the request to cite the following paper:

Michalis Vrigkas, Evangelia-Andriana Kourfalidou, Marina E. Plissiti and Christophoros Nikou, “FaceMask: A New Image Dataset for the Automated Identification of People Wearing Masks in the Wild.,” Sensors. 2022; 22(3):896. https://doi.org/10.3390/s22030896.

The FaceMask Database consists of 4,866 images of people in diffrerent environments and situations. These images were manually collected from Google Images and annotated with LabelImg tool in YOLO format. They are divided into two categories containing people wearing face masks and people not wearing face masks. The Dataset is splited into 3 folders -the training folder, the validation folder, and the test folder.
Categories

  1. Mask - Mask category is consisted of images depicting people wearing their mask correctly and people wearing their mask incorrectly, leaving their nose uncovered. There are 22,137 subjects in total wearing a face mask.
  2. No_Mask  - No_Mask category is consisted of images depicting people not wearing a mask at all, as well as people wearing a mask incorrectly without covering their mouth. There are 18,515 subjects in total not wearing a face mask. 
JavaScript Code

  1. Javascript code for downloading images and their URLs from the Web
Download JavaScript Code
FaceMAsk Dataset

  1. Training, Validation, and Test sets for the FaceMask dataset.
Download the FaceMask dataset

Description

Description of the FaceMaskDetector Dataset and its contents.

Extracted Features

Features extracted using the YOLOv3,YOLOv4 and YOLOv4-tiny.

Image URLs

Google Images urls

This work has been co-funded by the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship, and Innovation, under the call RESEARCH-CREATE-INNOVATE (Project Code: T1EDK-04517).
The authors gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research.