Mouth Masks Confuse Facial Recognition

Face masks are introduced in large parts of our country to limit the spread of COVID-19, but they also appear to have an effect on privacy.

 

Most facial recognition algorithms are not provided for the fabrics and paper patches, according to a study by the NIST, the American Institute for Standards and Technology.

Wearing face masks covering the mouth and nose, which is recommended in corona times, decreases the accuracy of most face recognition algorithms.

Depending on the algorithm, they register between five and fifty percent more errors.

In addition, black masks are more likely to provoke errors than blue ones, and the more of the nose covered, the more difficult it becomes for the algorithm to recognize the face.

The test was performed with algorithms trained before the pandemic, which explains this.

A later test should look at how good algorithms that have been specifically developed to recognize masked faces.

Leave a Reply