PFT delivers AI-led automation of Segmentation workflows to ITV
December 21, 2021
Prime Focus Technologies (PFT) has announced that its AI/ML-powered native media recognition engine CLEAR Vision Cloud has been engaged by ITV, the oldest and largest commercial terrestrial television network in the United Kingdom for delivering AI-based automatic segmentation of long-form content.
ITV currently marks about 12 types of segments in the content manually, which are both physical segments like color bars, blacks, and slates, and creative segments like recaps, PRTS, credits, program part segments, and bumpers.
ITV wanted to leverage AI/ML technologies to mark the segments at 95-100% accuracy followed by human QC, and complete the entire process in about 3 minutes instead of the 15-20 minutes it takes currently when performed manually.
PFT showcased CLEAR Vision Cloud and its AI-led Basic Metadata Data Pack and AI-assisted Segmentation Toolkit to ITV. When deployed, the data pack ensures near 100% accuracy and frame accuracy, and 100% automation of workflows to extract the content segments. With the segmentation toolkit, M&E organizations can review and QC automatically identified segments, filter out content segments and export EDLs, and generate a video of the custom segment, stripping out the rest of the physical segments.
“When we demonstrated CLEAR Vision Cloud and especially our segmentation use case perfected in-house by the Digital Media Services team delivering numerous client facing projects, ITV was convinced,” said Nick Kaimakami, EVP & Head – EMEA and USA East. “CLEAR Vision Cloud will bring about a substantial reduction in time and cost of segment marking and content segment extraction for ITV.”
ITV has deployed the CLEAR Vision Cloud, which automatically picks up the content and runs its sophisticated AI engines to mark the segments with near 100% detection accuracy and 95-98% frame accuracy, even in creative segments. Since CLEAR Vision Cloud is hosted on the cloud, it can auto-scale on demand and handle elastic loads at minimal costs while ensuring high performance always.
The AI-led segmentation toolkit that ITV operators will use has been custom-made by PFT specifically to ensure high performance at peak load and very low manual review time. PFT worked jointly and directly with the operators, architects and engineering teams inside ITV to learn the process and tune its AI engines to deliver the outcomes. A Segmentation proof of concept (POC) was done jointly and the accuracy was tuned up to the required level so that a QC could be done in less than 3 minutes.
“I am very impressed with CLEAR Vision Cloud and PFT’s approach to making AI work for us,” said Sonny Hanley, Controller of Content Services, ITV. “PFT understands that each media enterprise is unique both in terms of business objectives and nature of content, and hence the AI solution cannot have a ‘one size fits all’ approach. Their AI experts worked with our teams to understand the challenges and deliver a solution that truly meets the efficiency goals we had envisaged when we commissioned this project.”
A truly novel industry-first feature of CLEAR Vision Cloud is that when a content segment or its signature is not identified by its AI engine and is marked manually during QC, it automatically triggers learning for Vision Cloud’s Machine Wisdom. This automatic learning through QC input ensures errors are never repeated and frame accuracy improves every time. Machine Wisdom is like the human brain, constantly learning and applying those in context to deliver improved outcomes for the customer.