FakeCatcher, Intel’s Deepfake Detector Has 96% Accuracy

Deepfake is the new talk of the town these days. For those who don’t know about it it is a process where someone’s existing image or video is replaced with someone else’s. This process takes place digitally. For this Intel has launched a deepfake detection tech named FakeCatcher that can spot deepfake video 96% of the time.  FakeCatcher was developed by Intel researchers Ilke Demir and Umur Ciftci from the State University of New York at Binghamton using Intel tech.

The company revealed that Fakecatcher can operate in real-time to detect deep fake videos and also told that this is the first of its kind tool in the world to detect fake videos. Fakecatcher has a 96% success rate when it comes to detecting fake likenesses. As far as its mechanism is concerned, it collects data based on blood flow mechanics on a person’s face by scanning the pixel in a video. After this a deep learning AI can determine is the subject’s likeness is authentic or not.

While telling about Fake Checker, Intel Labs senior staff researcher Ilke 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 hosted on a server and interfaces with video by using a web-based platform. Its approach is quite different from the traditional deep-learning-based detectors, which detect the fake aspects of the video whereas FakeCatcher detects the real part of the video.

With the growing tech, deepfake technology has grown during the last few years. Even the FBI reported to  Internet Crime Complaint Center that it had received too many complaints of people who were using deepfakes in videos.

If Intel is true regarding the detection of deepfake videos through FakeCatcher, it’s a big step and will be able to lead people toward the right place.

Also Read: Bruce Willis Denies Selling Rights to Deepfake Company