Each February, Amsterdam hosts the largest AV and systems integration showcase in the world - Integrated Systems Europe (ISE). Here, the global AV industry gathers to celebrate leading AV projects as well as cutting-edge technologies, products and solutions.
At ISE 2020, we were showcasing our latest updates for the Pro AV and broadcast industry - a collection of newly available AI capabilities and the first implementation of HDMI 2.1 on our 7nm Versal ACAP devices.
AI/ML continues to gain precedence across all industries, and our capabilities generated a lot of interest at the show. Below is snapshot of the demo's we were giving.
Region-of-Interest (ROI) Encoding
Streaming and storage costs for video is of major importance to the broadcast industry. Region-of-Interest (ROI) encoding can help reduce costs by reducing the overall bitrate of content while maintaining video quality (VQ) in areas where the eye is naturally drawn - particularly faces and people, while reducing the VQ in less important areas such as backgrounds.
AI models automatically detect and track faces and dynamically apply window coordinates to the encoder to determine where lower QP (quantization parameters) can be applied to provide higher VQ, whilst the rest of the scene receives higher QP (lower bitrate).
Intelligent Digital Signage
Highly-targeted advertising is the holy grail for marketers. Using AI to analyze the audience in front of a digital sign makes it possible to serve more relevant and targeted ads, based on metrics like age and gender. This makes the signage provider more attractive to advertisers who will pay more for specific target audiences. This also generates valuable data for the advertiser such as viewer interest, which can also be monetarized.
Object Tracking & Windowing
Face detection using AI has many applications in the broadcast industry. For example, imagine live-streaming a panel discussion about an artist’s work at a local college. This is a low budget event with a niche audience, so production costs need to be low. Likely a single camera will be used, capturing the whole panel with occasional zooming and panning.
Using AI, it’s possible to have a static 4K camera capture the whole panel, but automatically create extra, lower-resolution HD windowed outputs around each of the panelists and track them through the conversation. Thus with a single 4K camera, it’s possible to have multiple output shots to switch between during the live stream – a wide angle and multiple close-ups. This creates more visual interest and doesn’t require any extra camera equipment to set up – the camera operator can become the video mixer and simply select which frames to stream.
This AI technique is also useful in video conferencing applications, enhancing the capabilities of digital zoom with automated tracking.
Speech recognition using natural language processing (NLP) AI models is another area that we demonstrated. This type of AI is already used by smart devices that can respond to voice commands. Whether the processing occurs in the cloud or at the edge, adaptive technology from Xilinx can provide a significant power reduction versus a CPU implementation.
We’re excited to add these new AI capabilities into Xilinx's adaptable platforms to provide customers in the Pro AV and broadcast markets with advanced capabilities for monetizing analytics, improving workflow efficiency and enhancing usability. Ultimately, these integrated AI features will allow companies to increase innovation, differentiate themselves and accelerate time-to-market. You can find out more about Xilinx AI solutions here.