Certified Human: How new Intel tech detects deepfakes in real time | Intel

Опубликовано 31 октября 2022, 15:00

Intel Senior Staff Research Scientist Ilke Demir explains the concept of deepfakes and how Intel's solution recognizes manipulated videos. Intel’s deepfake detection platform, using FakeCatcher [1] technology, identifies deepfakes with high levels of accuracy in real-time, a first for the industry. In addition, eye/gaze-based deepfake detection [2] and source GAN detection [3] are also developed by her team to aid in the battle against deepfakes.

While the purpose of these solutions is to catch harmful fakes, deepfake technology can also be used for good. Dr. Demir shares how responsible deepfakes can be created by their multi-source synthesis approach [4] combining regions from multiple faces into a non-existing face, or how it can be used to protect the identities of individuals under sensitive situations by quantifiably dissimilar deepfakes [5].

The system is built on the 3rd generation Intel® Xeon® Scalable server, which enables up to 72 concurrent detection streams thanks to Intel AI optimizations, such as OpenVINO™ AI models, Intel® Deep Learning Boost, and Intel® AVX-512, accelerating the artificial intelligence and media processing capabilities in the system. What’s more, it’s the first real-time deepfake detection platform!

Learn more about our deepfake detection solution here. intel.com/content/www/us/en/re...

[1] U. A. Ciftci, I. Demir I, L. Yin. “FakeCatcher: Detection of Synthetic Portrait Videos using Biological Signals”. IEEE Transactions on Pattern Analysis & Machine Intelligence. July 2020.
[2] I. Demir I, U. A. Ciftci. “Where Do Deep Fakes Look? Synthetic Face Detection via Gaze Tracking”. ACM Symposium on Eye Tracking Research and Applications. May 2021.
[3] U. A. Ciftci, I. Demir I, L. Yin. "How Do the Hearts of Deep Fakes Beat? Deep Fake Source Detection via Interpreting Residuals with Biological Signals." IEEE International Joint Conference on Biometrics (IJCB), Sept 2020.
[4] I. Demir I, U. A. Ciftci. “MixSyn: Learning Composition and Style for Multi-Source Image Synthesis”. arXiv preprint arXiv:2111.12705. Nov 2021.
[5] U. Ciftci, G. Yuksek, I. Demir. “My Face My Choice: Privacy Enhancing Deepfakes for Social Media Anonymization”, IEEE/CVF Winter Conference on Applications of Computer Vision. Jan 2023.

See more and discover how Intel sparks innovation and pushes the boundaries of possibility: intel.com/stories

Subscribe now to Intel on YouTube: intel.ly/3IX1bN2

About Intel:
Intel, the world leader in silicon innovation, develops technologies, products and initiatives to continually advance how people work and live. Founded in 1968 to build semiconductor memory products, Intel introduced the world's first microprocessor in 1971. This decade, our mission is to create and extend computing technology to connect and enrich the lives of every person on Earth.

Connect with Intel:
Visit Intel WEBSITE: intel.ly/Intel
Follow Intel on TWITTER: intel.ly/Twitter
Follow Intel on INSTAGRAM: intel.ly/Instagram
Follow Intel on LINKEDIN: intel.ly/LinkedIn
Follow Intel on TIKTOK: intel.ly/TikTok



Certified Human: How new Intel tech detects deepfakes in real time | Intel
youtube.com/intel