Linius Virtualises Video for Monetisation and Personalised Viewing
The best techniques for readily searching video and obtaining results that apply to an individual's specific interest, at any given moment, remain elusive. Visually, video is often more beautiful, immediate, fun and entertaining than text. But for many people, capturing all the details and knowing how to find them later in a precise way, is a challenge.
Verbal approaches to analysing video, like speech to text, word search and voice recognition, are improving rapidly now but may be difficult to refine and the output may be too generic and require still more manual searching. Face detection is also becoming faster and more accurate all the time, but even when it works perfectly, finding a face does not tell you the whole story you have in mind. Neither of these techniques, and other similar ones, are accessible to everyone, nor are they convenient or automated in a customisable way.
From Static to Interactive
Digital Media World spoke to CEO Chris Richardson at Linius Technologies about the company's work in gaining control over information held within video. Linius converts static video into agile, interactive, virtual data. Chris said, “Creating a truly virtual video file means you can access its digital content at a basic, unbiased level. The company's proprietary Video Virtualization Engine makes it possible to tag, index, parse, splice, manage and manipulate video data directly online, before delivering it to the end user, as it moves between source and screen.”
All actions occur automatically once processing begins. The Linius Engine works on ordinary video files, breaking them down into controllable blocks of data, which it then searches through, tags and categorises. Once all of the data is indexed, a virtual file is created, which typically occupies less than one per cent of the original file size depending on the specifications.
Users also have a chance to enrich the data. Timecoded metadata is as basic to the processing as it is to any log and search operation. Linius gathers some incidental metadata as it works but users can add their own or third-party metadata to the process, and also apply AI or other business rules to make any of the video elements searchable.
Searching and Program Playback
The virtualisation work Linius does up to this point makes it possible to search the video content for any object - across multiple sources, down to single frame level – and immediately reassemble it at the point of playback as a highly customised video program, viewable on web-enabled devices or platforms. The programs may be a series of clips or even single frames. Virtualisation and re-assembly does not affect video quality, and varying aspect ratios between sources are handled on the fly to produce a smooth, watchable result.
Virtualisation gives owners more control over their video. It also means many more consumers will think of using video in more specific, granular ways. Individuals can use Linius' embeddable Search Widget to search for objects across video sources, and immediately playback the content that matches their search results. Service providers can automatically integrate the search results into existing video workflows, or use it as a pool of resource material that will be delivered on demand in an infinite variety of streams to individuals, based on consumption preferences or behaviour.
“Linius has identified several broad use cases – news, sports, education, corporate communications, security and defence, and sports betting. However, when video data is made available in such a re-combinable format, each new customer tends to have a new application in mind,” Chris said. It can be easier to imagine the results of the virtualisation-and-search process by considering the applications.
Linius for Broadcast, Education and Business
News and sports broadcasters and publishers can potentially supply every viewer with a precisely personalised newsfeed or let viewers independently search video archives to create their own news experiences. Users can search on combinations of people, places, sports action and similar elements. These abilities can lead to new services and subscription, analysis, targeting and revenue models. Also, the small size of Linius’ virtualised video reduces the need to move full-resolution video files around the network.
In education, a large amount of time, money and effort is spent collecting, storing and presenting video content. Students also spend many hours searching through recorded lectures to find small pieces of information. Using Linius to search for and find objects with text, audio and facial recognition gives gives lecturers, tutors and students more control over their resources.
In terms of the education business, access to personalised video learning experiences can make an institution more competitive and increase student engagement, and reduce video production costs at the same time. It is also necessary now to publish research results, but because detailed analysis and comparison of data points is usually slow and prone to error, being able to automatically count, analyse and compare elements within video content is valuable. www.linius.com