It is extremely difficult to keep up with the development of deepfake technology. Almost every week there is some hair-raising news about this: most recently Tom Cruise was made a TikTok star, and a mom used deepfake to kick her daughter’s rivals out of school.
And after it turned out that eyes could knock down artificially generated profiles, researchers at Binghamton University came up with a solution that could deal a severe blow to counterfeiters by recognizing camouflage videos with 90 percent confidence. Scientists who have worked with Intel computer experts base their recognition process on changes caused by the pulse.
A device called FakeCatcher looks at the skin color, and watch for the small changes that occur as a result of the heartbeat. The technology uses a process called photopletymography, PPG for short, which also measures heart rate in various wearable devices such as the Apple Watch. The system checks the consistency of PPG signals from different parts of the facial skin: there is no such thing in deepfake videos, as the forgery program does not receive such information. In real videos, on the other hand, blood flows in the same way on both sides of the face.
The team, with the help of Intel, has built an extremely complex physiological data recording laboratory for the developments , in which 18 video cameras are nailed to the volunteers’ faces, as well as 3D images and thermal imaging, and the use of physiological sensors. A simple half-hour session generates enough data to take half a day to process
FakeCatcher technology is still in development: researchers say it’s not just deepfake that can be used. then it will deploy, but it will also help with telemedicine.
Hardware, software, tests, curiosities and colorful news from the IT world by clicking here!