Welcome to the Multi-Class Face Mask Detection System repository! This project aims to detect face masks in images and classify them into different classes based on the mask-wearing behavior. The system is designed to identify three classes: "Correctly Worn," "Incorrectly Worn," and "Not Worn."
In the context of the ongoing global pandemic, the importance of wearing face masks has been emphasized to prevent the spread of infectious diseases. This project provides a solution for automated face mask detection and classification, which can be useful in various scenarios such as public spaces, workplaces, and healthcare settings.
- Multi-class face mask detection: Correctly Worn, Incorrectly Worn, Not Worn.
- User-friendly interface for easy interaction.
- Supports image input for real-time and batch processing.
- Pre-trained model for quick deployment.
The model was trained on a carefully curated dataset, which includes images of individuals with different mask-wearing behaviors, annotated with corresponding labels. This dataset comprises 853 images with annotations in PASCAL VOC format, divided into three classes: ”With mask,” ”Without mask,” and ”Mask worn incorrectly.”
Preliminary results from the training process indicate an overall accuracy of approximately 69.89%. The confusion matrix from the model’s predictions revealed an uneven distribution of accuracy across classes, attributed to the imbalance in the dataset.