It tracks blood flow in the pixels.
Most people would probably say that deepfakes are fun: swapping one face with another – what's not to like? The technology is getting better every year, so you can even watch Terminator starring Sylvester Stallone, with a whole cast of different actors. However, the issue is exactly how real deepfakes start looking. While entertaining, impersonation can easily spread misinformation. So while some apps make fakes, others try to find them.
Intel introduced a real-time deepfake detector called FakeCatcher that can understand if the video is not real with a 96% accuracy rate. Developed as part of Intel's Responsible AI work, it is "the world’s first real-time deepfake detector that returns results in milliseconds."
"Deepfake videos are everywhere now. You have probably already seen them; videos of celebrities doing or saying things they never actually did," said Ilke Demir, senior staff research scientist in Intel Labs.
While most other detectors look at raw data, FakeCatcher looks for "authentic clues" in real videos, assessing what makes us human, like "blood flow" in the pixels of a video. These signals are collected from all over the face and then translated into spatiotemporal maps. After that, using deep learning, the researchers can detect if a video is real or fake.
The tech works on a web-based platform and can run up to 72 different detection streams simultaneously on 3rd Gen Intel Xeon scalable processors. Learn more about FakeCatcher here and don't forget to join our Reddit page and our Telegram channel, follow us on Instagram and Twitter, where we share breakdowns, the latest news, awesome artworks, and more.