Advanced online biometrics applications explored by Humans.ai, Digital Domain, Kakao

Several developers in the biometrics space have made new announcements related to advanced applications powered by artificial intelligence (AI). Romanian deep-tech startup Humans.ai partnered with Booming Jobs, Digital Domain unveiled ‘Zoey,’ an advanced autonomous human with facial recognition capabilities, and Kakao Brain announced the development of new face-swapping technology for the metaverse.
Humans.ai partners with Booming Jobs
Romanian deep-tech startup Humans.ai and its Software-as-a-Service (SaaS) synthetic video production platform Tovid.ai are partnering with Booming Jobs, a career platform in the Netherlands.
The collaboration will see Booming Jobs deploy Humans.ai’s AI and synthetic media technology within the recruiting process, to provide mass corporate communications presentations as well as individual interaction with AI-enabled virtual assistants.
“The challenge we have here is that we have tens of thousands of vacancies online and creating spoken and personalized videos for all of them would take us months and a large budget,” explains Booming Jobs Founder Jeroen Fikkers.
“But by using the AI technology of Tovid.ai, we see this problem solved and we are able to scale our content production drastically.”
In addition, the technology could also be used to change people’s appearance to avoid bias by creating neutral online avatars to avoid gender, race, age, and other types of discrimination.
Digital Domain unveils “autonomous human” Zoey
VFX studio Digital Domain has announced at the FMX conference in Stuttgart a new “autonomous human” called Zoey.
The virtual assistant is powered by machine learning, and can engage in conversations with multiple participants at once, access the internet to answer questions, and more.
The autonomous human is based on actress Zoey Moses, who worked with Digital Domain to create a set of facial movements, mannerisms, and a range of emotive expressions for the virtual model.
The digital Zoey also features various biometric capabilities. Specifically, it comes with built-in facial recognition software, which allows it to remember people, and AI-powered text-to-speech technology from WellSaid Labs.
Digital Domain confirmed Zoey will be available to license in the near future.
Kakao Brain upgrades ‘Face-Swap’ technology
Kakao Brain has announced the development of a new biometrics-based ‘face-swapping’ technology to give people metaverse avatars that more closely resemble them.
Dubbed ‘Smooth-Swap,’ the new solution is designed to expand the capabilities of face swapping through enhanced identity embedding.
The term refers to a vector representation of a face image that is used to compare identities. If the representation vectors of two faces are close enough, their identities are considered the same.
‘Smooth-Swap’ is trained via supervised contrastive loss, and is able to acquire its stable identity gradient by learning embedding with a higher smoothness. The improvements eliminate the earlier model’s weakness of adding handcrafted components and 3D face modeling that ultimately complicated its design.
To fix these issues, Kakao Brain equipped ‘Smooth-Swap’ with a simple U-Net-based architecture with an integrated smooth identity embedder intended to deliver cutting-edge performance.
“We are proud and excited to unveil the groundbreaking face-swapping technology, ‘Smooth-Swap,’ to the world,” commented Kakao Brain CEO Kim Il-doo.
“I strongly believe this technology will accelerate innovation in the face-swapping sphere, bringing us another step closer to the incredibly immersive metaverse we always dreamed of as well as the digital human services of the future.”
Kakao Brain will present a paper detailing the new technology at the Computer Vision and Pattern Recognition Conference (CVPR), taking place in Louisiana between June 19 and 24.
Article Topics
AI | biometrics | Digital Domain | digital identity | face swap | Humans | Kakao Brain | metaverse | research and development | synthetic data | virtual human
Comments