클라우드 및 IP 기반 실시간 비디오 솔루션 리더인 TVU Networks는 TVU Anywhere 실시간 IP 비디오 스트리밍 앱 개발을 위한 무료 TVU Anywhere SDK를 발표했다. ...
By Paul Shen, TVU Networks
October 26th, 2020
With just about two weeks to go (as of this writing) before Election Day, the need has never been greater for TV news operations to take advantage of the latest technology to assist them in reporting on what voters need to know.
One major TV station group has recently begun doing just that, leveraging our TVU MediaMind engine to apply artificial intelligence algorithms to pool feed footage. Thanks to AI, their journalists can sift through thousands of hours of footage collected every week and find the content they need to do their stories.
The workflow benefits are staggering. Being able to find just the right footage from 5,040 hours of content per week –that’s 30 camera feeds, 24 hours a day, seven days a week—would otherwise be impossible. But with AI to create metadata for each frame of video, every bit of the floor speeches, the committee debates, impromptu press conferences, campaign events and confirmation hearings is accessible.
The AI Advantage
Our TVU MediaMind AI engine employs a variety of algorithms, including speech-to-text, facial recognition and object recognition, to create metadata at the frame level that makes reporters’ searches for the right content fast and easy.
Recently, however, there’s been an upgrade of sorts to the object recognition algorithm. Rather than simply identifying the large items in a frame –a building, a truck or an airplane like a Boeing 747—the algorithm has been fine-tuned, so to speak, to be able to create metadata for each small element in a frame.
So, rather than simply recognizing that the airplane in a frame is a 747, the algorithm now recognizes there are bleachers full of people in front of plane, a stage, a podium and a blond-haired man in a trench coat standing on the stage, for example.
The significance is that rather than retrieving all of the clips with President Trump or a 747 or fans in bleachers, the metadata created about all of the objects in each frame enables a journalist to hone in on exactly what is needed for the story.
What this really boils down to is an AI-enhanced workflow. For the reporter in the trenches, an AI-enhanced workflow means far less time spent on tedious tasks like scanning through hours of footage and transcribing audio and far more time spent being productive.
For the station group that has deployed TVU MediaMind, the AI-enhanced workflow means it’s stories are being made better by use of more relevant clips, more stories are being created to fill its linear, digital and social media pipeline and ultimately greater viewer loyalty will be built, leading to better ratings and higher potential revenue.
Ultimately, imagine what this or any other large news organization could do if all of the footage its news photographers capture was enhanced by the addition of AI-generated metadata. Think of how much better the news production workflow at all of its stations could become and how much better the news content presented to viewers would be.
For more information on this topic or to contact the blog authorContact us