ARTIFICIAL INTELLIGENCE
72
AI first has to be consolidated in performance venues, and then as a second step could be ported into a stadium”
AI adopter Naostage’s K System is claimed to be the only beaconless 3D tracking system on the market
Ronja Harste, Sennheiser’s portfolio manager for Business Communications, recalls: “During the pandemic, the need for improving hybrid meeting experiences became much higher as attendees wanted to have similar experiences remotely to what they would experience in face-to-face meetings. Whereas audio is still the most important part of being able to understand each other clearly, advanced video quality and algorithms became significantly more important.” With increased meeting equity becoming a near-universal goal – including for those participants in the room who are ‘sitting far away from the camera and can often be hard to see for the remote participants’ – Sennheiser integrated AI-based camera features into its Team Connect Bar Solutions family of scalable, all-in-one conferencing devices with built-in camera, microphones and speakers. AI-based features such as auto framing and person tiling – which allows users to view multiple participants at once during a video call – enable remote attendees to view the movement and gestures of people more clearly within the meeting room. Harste is not alone in identifying the ability of AI algorithms to improve video experiences and, on the audio front, filter out unwanted noises. But, there is also a note of caution as tracking technology becomes more ubiquitous: “As people and voices are tracked, often recorded and sometimes even assigned to individual voice profiles, security will become a much bigger topic to ensure the protection of personal data.” TRACKING TECHNOLOGIES Staging and performance-related applications are another growth area for AI in AV. Among the companies leading the way here is Naostage, whose K System is described as the world’s first and only beaconless 3D tracking system. It comprises three elements: Kapta, a lightweight 3D sensor bar that scans in real time; Kore, an IP computing unit – trained by deep learning over thousands of show hours – that merges sensor data and automatically detects, tracks and identifies people moving within the tracking areas; and Kratos, a complete user software to design projects and manage interactions with external AVL systems. Invited to consider Naostage’s starting point for its exploration of AI, CTO Olivier Le Doeuff responds: “This is a tough question. The simple answer is that it has become a de facto method used in the computer vision industry. Our AI models are essentially extensions of the previous algorithm we were using, but instead of manually tuning all the parameters of the algorithm, we rely on a so-called training step. It sounds ‘magical’ and gives the impression that the model
Is experia solut est ex eum num facium, sed quunt, ut aut eum am natur?Verumet quia veris am, Companies such as Naostage are implementing AI to enhance staging and performance-related applications industry-wide
Powered by FlippingBook