This Face Recognition Tech Can See Through Niqab! Privacy Breach?

Face Recognition technology may detect a person's face through Burqa!

It is critical for security to precisely identify persons (and not embarrass yourself by hugging strangers). It was given as the primary justification for barring and regulating those wearing religious heads and facial coverings in public settings.

A plan to force Muslim women who wear face coverings or niqab to sit in glassed-in enclosures at Parliament House has been scrapped, but the issue remains: how proficient are we at recognizing people based on their facial features?

Face recognition systems employ computer algorithms to recognize identifiable aspects of a person's face. The data is then translated into a mathematical representation and compared to data from other faces in a face recognition database. These traits include the space between the eyes and the curve of the chin.

Face-recognition technology is rapidly evolving and is being utilized in a variety of industries such as marketing, education, criminal investigation, security, and biometrics. It can now assess the individual's facial expression in addition to identifying them. The limitations of facial recognition software when a person's face is partially obscured, as might occur while wearing a veil or protective face mask, are the focus of a study published in the International Journal of Computer Vision.

A substantial amount of study has been dedicated to full-face biometric identification. However, using faces that are only partially visible, such as veiled people, is tricky. In this study, the deep convolutional neural network (CNN) is used to extract properties from images of veiled people's faces.

DeepVeil, the team's proof of concept, employed face-on photographs of veiled persons captured up close in conjunction with an internal picture database. The next step will involve working with a more diverse collection of photographs collected in various circumstances, including images taken from various angles. However, as algorithms and software have evolved, a clean face-on image is no longer required to establish a person's identity, as it was in the early days of the standard facial recognition system. As a result, DeepVeil may go through the same thing with the right approach and continuing development.

Identification and facial characteristics

Coverings that hide the majority of the face, such as the burqa, are going to make identification difficult. However, coverings like the headscarf or hijab tend to obscure outward traits such as facial shape, ears, and hairline.

People's ability to identify two photos of a face when just internal facial characteristics are presented, as in these images, has been compared to when all features are revealed in studies. The results indicate that when only internal data are present, recognition accuracy improves; however, the existence of exterior features might impair identification. To facilitate identification, perhaps we should all wear headscarves.

External aspects of the face, notably the hairline, can quickly be modified and hence may not serve as a reliable foundation for identification. Importantly, a new study reveals that faces shown with a headscarf are perceived as more similar than those without. This implies that the headscarf affects people's impressions of faces even when they are instructed to focus solely on the interior aspects of the face.

Differences in face recognition across cultures

Researchers Ahmed Megreya and Markus Bindemann discovered discrepancies in performance between Westerners (British University students) and persons from Middle Eastern cultures (they looked at Egyptians). People from cultures where head coverings are widespread performed better at recognizing people with and without the covers.

This effect has been related to the cross-race effect, in which people recognize people of their race better. Both effects show that our experiences have a major influence on our sense of identity. The outcomes of this study show that we may be able to improve our ability to recognize people based on their faces. The tricky part is determining how.

Other factors have a larger impact

What about severe changes in appearance caused by aging, weight loss or gain, or plastic surgery?

Age-related changes in facial appearance are especially difficult to identify. Children who are the subject of custody orders or who have gone missing might easily go undiscovered by authorities since the photographs they have rarely match their looks.

Despite popular portrayals of forensic artists creating appropriate "aged" images for identification, relatively little scientific study into the accuracy of this and other kinds of image-based identification has been conducted. At present, studying the effects of aging on identification accuracy is a top focus. Passport officials are taught to look for weight loss or growth, as well as plastic surgery, as possible sources of identity mistakes. However, we lack data to establish if this training is successful.