A new software developed at the University of Cambridge in the UK, opens up new possibilities in facial recognition. It can see people wearing hats and scarves. This machine learning algorithm can recognize 14 key facial features that the brain pays attention when we look at someone’s face.
If a person is wearing a hat, glasses or a fake beard, the software can now tell you which of them it is in as little as 14 guesses. The algorithm examined photos to identify individuals disguised in a photo. It then scanned other images of the person, looking for more examples. This system was able to identify 3 out of 4 people wearing a headscarf.
The success rate with these percentages may seem lower, but Singh points out that it is the best so far in a variety of different scenarios related to identifying disguised faces. The software only needs to see a small percentage of key facial features, which includes many around the eyes and mouth. “In essence, the system is able to see through your mask,” Singh told New Scientist who added that “AI works very well on these kind of disguises, since it processes the key parts of the face.”
The system could recieve further developement and patches to recognise people who wear masks on their face. The algorithm helps one identify criminals escaping from the law. While it’s true that the system could be used to track down protesters and dissidents, Singh admits that it has an upside too. It has many important benefits like saving time and effort, increasing quality and reducing the price..” he said.
You can read another article about software application here.
Author: PC-GR
The World of Technology
