baskoroadiw/maskedfacesegmentation-dataset

Datasets for multiclass facemask semantic segmentation and classification

MaskedFaceSegmentation-Dataset

Welcome to MaskedFaceSegmentation-Dataset repository

This is dataset for multiclass facemask semantic segmentation and classification. Dataset contains face images that use masks correctly and incorrectly. The image is obtained from MaskedFace-Net dataset [1].

There are 500 images which are divided into 2 (two) classes:

• 300 images with Correct Wear Mask (CWM) at 1024x1024
• 200 images with Incorrect Wear Mask (IWM) at 1024x1024

The dataset is also divided into training data and testing data :

• 450 images for training data
• 50 images for testing data

Below is a summary of data distribution from the dataset :

Class Name	Data Training	Data Testing	Total
CWM	275	25	300
IWM	175	25	200
Total	450	50	500

Example Images and Ground Truth

References

1. Cabani, Adnane, et al. "MaskedFace-Net–A dataset of correctly/incorrectly masked face images in the context of COVID-19." Smart Health 19 (2021): 100144. -- Link