Kakao Brain Brings Many Advances To The Metaverse With New Face-Swap Technology

Representing another milestone in Kakao Brain’s face-swapping research, the company’s paper “Smooth-Swap: A Simple Enhancement for Face-Swapping with Smoothness” will be presented at the upcoming World Vision Conference. by computer, CVPR 2022[2], for the second consecutive year. This will include an exclusive oral presentation session reserved for the most outstanding of the papers received (25.33% of the 8,161 submissions were received this year). At last year’s event, only 4% of accepted papers had time for an oral presentation in which Kakao Brain was nominated for its outstanding research role, “HOTR: End-to-End Human-Object Interaction Detection with Transformers “. This year, ‘Smooth-Swap’ has not only succeeded in significantly reducing the complexity of its architecture, but also has great market potential, both recognized and presented at the first computer vision conference.

An accurate and consistent birth gradient[3] essential to change the identity of a person without sacrificing high image quality. Trained through supervised contrastive loss, ‘Smooth-Swap’ captures a strong birth gradient by learning to include more smoothly. These improvements address the weakness of the previous model by adding hand-made components and 3D surface modeling, which ultimately complicated its design and resulted in sophisticated hyperparameter tuning. However, “Smooth-Swap” relies on a simple U-Net-based architecture with a built-in identity integrator to provide industry-leading performance.

The simple architecture and improved performance of “Smooth-Swap” not only makes the technology competitive in terms of market potential and wider application, but it is also able to cope with face-to-face swapping scenarios. . “Smooth-Swap” suggests a different way of combining identities and allows the generator to create higher quality images, especially if the face shape of a subject is changed. Thanks to Kakao Brain’s “Smooth-Swap”, which enables fast and robust face-to-face exchanges, it is expected to develop different types of digital people such as virtual influencers, show hosts and announcers.

“We are proud and excited to open up the revolutionary face swapping technology,‘ Smooth-Swap ’, to the world,” said Kim Il-doo, CEO of Kakao Brain. “I firmly believe that this technology can accelerate the face-to-face innovation, bringing us closer to the incredible immersive metaverse we have always dreamed of, and the digital human service of the future.”

About Kakao Brain

Kakao Brain is a world -leading AI company with unmatched AI technologies and research and development networks. The company was founded by Kakao in 2017 to solve some of the most “unthinkable questions” in the world with solutions made possible through life-changing artificial intelligence technologies. Always at the forefront of innovation in the world of technology, Kakao Brain has developed many revolutionary AI services and models designed to improve the quality of life of thousands of people, including minDALL-E, KoGPT, CLIP /ALIGN and RQ-Transformer. As a global AI pioneer, Kakao Brain has a responsibility to develop a vibrant technology community and a strong R&D ecosystem as part of its mission to create a new technology market with endless potential. potential. For more information, visit https://KakaoBrain.com/.

[1] Identity aggregation is a vector representation of the face image used to compare identities. If the representation vectors (or integration vectors) of the two faces are close enough, their identities are considered identical.

[2] The CVPR (Conference on Computer Vision and Pattern Recognition) organized by the Institute of Electrical and Electronics Engineers (IEEE) and the Computer Vision Foundation (CVF) since 1983, is considered one of the most recognized annual conferences in the vision sector. in computers, with the European Conference on Computer Vision (ECCV) and the International Conference on Computer Vision (ICCV).

[3] The identity gradient is a training signal that tells the face exchange model which part needs to be tuned to accurately transform the person’s identity.

SOURCE Cocoa Brain

Leave a Comment