2 min Analytics

AWS: Face recognition software Recognition now also recognizes fear

AWS: Face recognition software Recognition now also recognizes fear

Amazon Web Services’ (AWS) facial recognition software Recognition has been updated to recognize a person’s fear. The facial analysis function is also more accurate and functionality has been improved.

According to AWS itself, the software could already accurately recognize seven emotions. It’s about cheerful, sad, angry, surprised, disgusted, calm and confused. An eighth emotion has been added with fear, writes IT Pro.

However, the question is how accurate the recognition of emotions really is. Some experts argue that there is scientific evidence to suggest that there is a correlation between facial expressions and emotions, but that the ways in which they are communicated vary from culture to culture. Situations can also be very different.

AWS further states that the software is now more accurate about recognizing the gender of the user. Recognition can also recognise the age category of the subject, as well as attributes such as laughter and the quality of an image of a face.

Ethical Use Recognition

Recognition has been the subject of criticism and concern for some time now, especially when it comes to the ethical use of facial recognition software. In April, for example, a group of more than twenty artificial intelligence investigators asked AWS to stop selling the software to the police.

According to the group, there are no rules or restrictions to prevent abuse. There are no laws or mandatory standards to ensure that Recognition is used in such a way that it does not violate civil rights.

Recognition is used by the police in Florida and Washington. In addition, it would have been offered to the U.S. Department of Homeland Security to identify immigrants.

Software has prejudices

Moreover, research by MIT in January showed that the software seems to be biased when it comes to women with a dark complexion. 31 percent of women with dark skins were wrongly classified as men. On average, 19 percent of women were identified as men.

This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.