New algorithm can outsmart humans in recognising faces
Researchers have developed a new algorithm that has for the first time allowed a computer to outperform humans when attempting to distinguish between faces.
Washington: Researchers have developed a new algorithm that has for the first time allowed a computer to outperform humans when attempting to distinguish between faces.
Chinese scientists fine-tuned an algorithm used for distinguishing between faces in photographs and tested their system, called GuassainFace, on a database known as Labelled Faces in the Wild (LFW).
The LFW is a database with a dataset of 13,000 headshots of famous people and has become a standard benchmark for testing facial recognition using a computer, `phys.Org` reported.
In the study, GuassainFace results showed a success rate of 98.52 per cent on the LFW, compared to an average of 97.53 per cent for humans - the first time a computer has ever beaten the human average, researchers said.
Researchers said that people get better at recognising faces the more often they see them. It clearly has something to do with adding more data as a person is seen from more angles, in different light, using different expressions, while wearing makeup or not, etc.
The same thing is true for a computer. In order to discern if two photos show the same person, the computer has to have seen that person before in multiple environments.
To allow that to happen, Chaochao Lu and Xiaoou Tang, from Chinese University of Hong Kong, exposed their system to multiple datasets, such as the Multi-PIE database or Life Photos.
Both offer multiple pictures of the same person to allow for not just better comparison capabilities for those in those datasets, but for getting smarter in general regarding how to match faces in other datasets.
Beating humans on the LFW is a remarkable achievement, of course, but it is just one benchmark, researchers noted, adding that computer technology still has a long way to go before matching the abilities of humans in generalised surroundings.