A new study recently suggested that up to 30 percent of humans are fooled by the faces they see around, yet artificial intelligence or AI is getting it right all the time.
As indicated in a Mail Online report, at first glance, such faces appear like real people. However, many of these faces have been spoofed, and they include staged pictures, cutouts, and masks.
The study tested both machines and humans by providing them with the most common spoofing method such as printed photos, digital images, videos, and 2D or 3D masks.
The researchers explained that computers were more precise with each type of image, scoring zero percent error rates throughout all 175,000 images and all attack types.
Humans Vs AI
Humans were found to have an extremely lower degree of preciseness for each spoofing technique type which includes misidentifying 30 percent of photo prints which is one of the simplest types of attack for fraudsters to implement.
Con artists frequently try to imitate actual customers during processes like creating a new bank account or logging into a currently existing account.
Even when a 17-member group voted on the images leading to a more accurate result compared to a person, their majority decisions were never better compared to the performance of the computer for the same task.
On the other hand, AI was found to be nearly 10 times faster to recognize a picture of an actual or a spoofed face.
According to a similar The World News report, on average, it took humans about 4.8 seconds for each image to identify liveness, while computers that run on a single CPU took shorter than 0.5 seconds.
Actual Faces Vs Spoofs
The new study, Human or Machine: AI Proves Best at Spotting Biometric Attacks, proposed that computers are developed more proficient compared to people. They can precisely and quickly identify if a photo is a precise, liver person against a presentation attack.
In the said research, the AI system mistakenly categorized just one percent of actual faces as spoofs. As for humans, they misclassified 18 percent of actual faces as spoofs.
These latest advances in technology support the rapid increase in facial recognition for determining verification and authentication, the New York-based company ID R&D which commissioned this study said.
According to the company, the findings also provided robust evidence for organizations in financial services, as well as other industries that stake trust in automation.
AI Facial Liveness Technology
In a related report, VN Explorer said that the ability to use AI facial liveness technology to detect fraud saves time and enables human resources to focus on more complex fraud, ID R&D said.
The company said the research verifies that "passive facial liveness detection" which instantaneously validates if photos captured in real-time are of a live person, is better than humans at keeping real customers out of the fraud net, as well.
Alexey Khitrov, CEO at ID R&D which provides AI-based face and voice biometrics and liveness detection technologies said the findings are undeniable.
He added that biometric technology used for the verification of identity has evolved in recent years to enhance speed and preciseness and that this is now substantially outperforming the human eye.
Companies can achieve efficiencies by applying identity verification systems including a biometric component. Nonetheless, there is still work that needs to be done. Khitrov said that they are excited to see biometrics that helps build "consumer trust."
Related information about human face and 3D mask is shown on the University of York's YouTube video below:
RELATED ARTICLE : Pantyhose as COVID-19 Protection? Study Reveals How It Helps Shield Users from the Virus More Effectively
Check out more news and information on Face Masks and Artificial Intelligence on Science Times.
