Detection of human faces covered with disguise and makeup

Abstract EN

Face detection is kind of the identification.

When we look at someone's face, we can get information like his or her gender and age.

Face detection research has exploded in popularity during the last few decades.

Starting with algorithms that can detect faces in constrained environments, today's face detection systems can attain extremely great accuracies at the large scale unconstrained facial datasets.

While new algorithms continue to increase performance, the majority of face detection systems are vulnerable to failure when subjected to disguise and cosmetics alterations, which is one of the most difficult covariates to overcome.

In this article, the database of disguised and makeup faces (DMFD) is employed.

In order to address this issue, we detected the location and size of the facial in the image by using Histogram of Oriented Gradients (HOG) + Linear SVM Machine Learning detector on the Disguise and makeup face database (DMFD).This approach is effective and can detect any disguise and makeup faces in the complex background and illumination variation.

The results shows the effectiveness of the face detection system on a database (DMFD) and it provided better results of (99.3%).