To determine the sexual orientation of people from facial images or photographs, the researchers have used a novel technology. With more successful detection than people.
The study is credited to Yilun Wang and Michal Kosinski and the details of the study are to be published in the Journal of Personality and Social Psychology.
The study from Stanford University is presently available in a draft form. Authors from Stanford University, the USA entitled their study, “Deep neural networks are more accurate than humans at detecting sexual orientation from facial images.”
This study shows that faces are more informative about sexual orientation than people previously thought. Deep neural networks is a form of technology that extracts some certain facial features from the facial images.
The computer assesses the facial features and detects if the person is hetero or homosexual.
For the study, the team included 130,741 images of 36,630 men and 170,360 images of 38,593 women obtained from an American dating website having public profiles.
To check for all those with most clarity, all the images were first detected using a facial-detection technology. This left 35,326 pictures of 14,776 people with a fair representation of both homosexual and heterosexual men and women.
VGG-FACE technology was used to get first detected the face. Later this gave out “faceprint” -a string of numbers that characterize the face. At the end, the technology was analyzed using statistical tests to detect the sexual orientation of a person.
For each image, it was successful in discriminating homo and heterosexual men in 81 percent of cases and between homo and heterosexual women in 74 percent cases.
Compare this to the guess of humans which scored 61 percent for men and 54% for women.
The study results revealed the accuracy of the algorithm used to the technology accurate to 91% and 83% when five facial images of the same person were fed to it.
This study included two types of facial features:
Fixed type: the same shape of the nose or other parts of the face that are unalterable.
Transient type: with grooming style that could change According to the wishes of the individual.
The previous studies, in this case, have shown the deeply rooted basis of the sexual orientation like the hormonal makeup of the individual. Facial morphology or shape even expressions and grooming styles are determined by hormones.
The algorithms showed 57 percent accuracy in case of men and in case of lesbian females, it was 58 percent accurate.
Shapes of the noses, eyes, eyebrows, cheeks, chin, and hairline were noteworthy features in case of male sexuality. While the female sexuality was detected taking into account shapes of the nose, corners of the mouth, hair, and neckline.
Dr. Kosinski gives his perceptions that this type of AI systems can also detect other intimate and private information such as IQ or political views of a person.
Dr. Kosinski says that humans have missed a lot of information with their naked eyes that the machines would not. He is well known for his previous work on psychometric profiling using information from social networking sites such as Facebook to draw conclusions regarding the personality of some male or female.
But I think the accurate detection of gay persons could be the potential threat to the privacy and safety of the gay men and women in future.
Both Donald Trump’s campaign and Brexit supporters have used this technology to target voters.
The technology that has raised concerns about what is private and personal and what is not, could be used in the organizations where recognition of male or female is necessary.