Mask Detection

General Info

This program is to detect faces and classify whether people are wearing mask properly. There is three categories:

• mask: if the person is wearing mask properly.
• incorrect_mask: if the person is not wearing mask properly.
• no_mask: if the person is not wearing mask at all.

Despite the fact that I apply deep learning, the software can operate at realtime speed. The essential components are detection and classification. However, this method has a problem because it takes longer to process when more people are spotted in a frame. When it identifies more than one individual, the FPS may decrease. Regardless of the disadvantage of the pipeline, this prototype provides a very good detection result. Please see my github repository for further information.

Output

Below is an example of the detection result.