Nvidia: AI Could Solve Video Call Issues

AI the solution to Video Call problems?

3
Nvidia-ai-solve-video-problems
Collaboration

Published: October 15, 2020

Rebekah Carter - Writer

Rebekah Carter

Video calling isn’t just a communication option anymore – it’s an essential part of the collaboration stack. Ever since the rise of COVID-19 at the beginning of 2020, the adoption of video is skyrocketing. Microsoft even hit a new daily record on Teams, of 2.7 billion meeting minutes in a day in April. Of course, there are still challenges to overcome in the video-first landscape.

On one hand, video calling allows for more immersive face-to-face conversations with your colleagues. It adds context and humanity to a meeting in a way that mimics in-person interaction. However, issues like face alignment problems and excess noise can detract from the quality of a meeting. Fortunately, Nvidia, a leader in computing technology, thinks it may have the answer.

Nvidia recently announced a new videoconferencing solution aimed at developers. With Nvidia Maxine, the company believes that developers can fix problems in video calls using the latest artificial intelligence technology.

What is Nvidia Maxine?

Nvidia Maxine processes calls in the cloud using the Nvidia GPUs that are currently leading the market in performance. Maxine also boosts the quality of video conferencing conversations in a variety of ways, thanks to artificial intelligence support. Intelligent technology ensures that Maxine can realign faces and gazes in a meeting, so they seem like they’re looking at a camera directly.

At the same time, the service reduces the requirements for video bandwidth, down to a tenth of the requirements generally needed for H.264 streaming compression. The service does this by transmitting key facial points only, and upscaling the resolution of videos. Other features available here include re-lighting for user faces, real-time transcription and translation, and even animated avatars.

Not all of the features listed for Maxine are brand-new to the marketplace. Real-time transcription and video compression are common solutions that appear in various video conferencing tools. Apple and Microsoft also match Nvidia by offering gaze alignment functionality, helping people to maintain eye contact during video calls. However, Nvidia seems to be taking this functionality to another level.

Nvidia is also relying on its presence in the cloud computing industry, alongside its impressive work in resources and development for AI to help it make an impact on the landscape. What most companies will be waiting for is evidence that established videoconferencing brands will actually adopt the new Nvidia technology. After all, Maxine is a toolkit that supports third-party firms in enhancing their software. It’s not a consumer platform on its own.

At this point, Nvidia has only announced a single partnership with Avaya, which has agreed to use certain features from Maxine. However, all major cloud vendors are also offering Maxine as part of the Nvidia GPU cloud services too.

Smarter Video Conferencing for Everyone

Richard Kerris, the General Manager for entertainment and media at Nvidia, said that Maxine is an exciting and timely announcement for the company. The business notes that today’s companies are moving through a transformative era right now. We’ve all experienced the headaches of bandwidth limitations preventing the quality of a video conferencing session.

avaya logoApplying AI to the problems that teams face with their video meetings every day could be the perfect way to transform and improve the way that we meet. Nvidia believes that reconstructing different scenes with AI and transmitting only the data that needs transmitting will be a great way to reduce bandwidth requirements.

Nvidia also says that the compression feature in its solution uses an AI method called generative adversial networks or GANs to reconstruct faces in the cloud. This is the same technique currently used in many deepfake technology options. Instead of streaming a full screen of pixels, the software finds the most important facial points of each person. Doing this, the AI can then re-animate the face intelligently in the video on the other side. This makes it possible to stream video with less data moving back and forth over the internet.

As usual with these announcements, it will be important to see more of this tech in action and wait for any deals that Nvidia may choose to make with other companies that could heighten the potential of the technology. However, for now, Nvidia’s investment in the AI environment for video conferencing says some important things about the workplace, and where it’s heading.

The future of videoconferencing will be smarter than ever, with AI that can structure your face and straighten your gaze to reduce the issues that most companies have with bandwidth.

 

 

Artificial IntelligenceVideo Conferencing
Featured

Share This Post