Intel’s Latest Deepfake Detector FakeCatcher to Analyse Subtle Blood Flow in Videos: Details

FakeCatcher, a real-time detector of deepfakes, has been introduced by Intel. The company has claimed that the FakeCatcher produces an accuracy of 96 percent. The deepfake detector is expected to work by analysing the subtle blood flow in video pixels. It is designed using Intel hardware and software. FakeCatcher runs on a server while the interfacing is done through a Web-based platform. Ilke Demir, a senior staff research scientist in Intel Labs, along with Umur Ciftci from the State University of New York at Binghamton worked together to develop the detector.

Intel has introduced a real-time deepfake detector and named it FakeCatcher. As mentioned earlier, the FakeCatcher is expected to detect fakes by analysing the subtle blood flow in video pixels. This will make it different from other deep learning-based deepfake detectors which look at raw data to try to find signs of inauthenticity and identify what is wrong with a video.

“When our hearts pump blood, our veins change colour. These blood flow signals are collected from all over the face and algorithms translate these signals into spatiotemporal maps. Then, using deep learning, we can instantly detect whether a video is real or fake”, the company explains the functioning of FakeCatcher in its blog post.

The deepfake detector was designed by Demir and Ciftci. Elaborating on the issue of deepfakes, Demir said, “Deepfake videos are everywhere now. You have probably already seen them; videos of celebrities doing or saying things they never actually did.”

FakeCatcher is a product of Intel’s Responsible AI work. Intel claims that the detector can catch fake videos with a 96 percent accuracy rate.

Deepfake has been a growing threat since the last few years. A study conducted by Forrester Research in 2020 had said that costs due to deepfake scams would exceed $250 million (roughly Rs. 2,040 crore) in that year.