3 Screen Solutions Applies Machine Learning with Media Distillery to Content Metadata Creation

Germany’s Android set top box (STB) software provider 3 Screen Solutions (3SS) is integrating machine learning technology for metadata creation from Media Distillery to enable deeper content search on the basis of audio or visual objects down to the frame level.

The technology will be combined with 3SS’ 3READY platform, which comprises a front end and User Interface.

It applies Convolutional Neural Networks, one of the principal variants of machine learning, to index content on the basis of components such as faces, single keywords, topics, logos and whole text strings, to create metadata in the first place. Then the 3SS platform would apply this in various ways to enhance content search and recommendation, so that for example users could be sent notifications when their favorite athletes, politicians or pop stars are on a linear show that is about to go out. Users can also search for content directly on the basis of such objects, or by context.

As Roland Sars, CEO of Amsterdam based Media Distillery, hinted, his company is aiming to solve the problem of audio/video search which has escalated in recent years because of the proliferation of content and also fragmentation of sources. “While it may be easy to search for content, we all know it is not always easy to find it,” said Roland Sars, CEO of Media Distillery. “With our new joint solution with 3SS, viewers will get much more personal TV experiences, and finding their desired content will be quicker, easier, and more intuitive.”

Such at least is the ambition and it remains to be seen how successfully that will be realized at a time when other vendors are piling into this field. It is certainly true that Media Distillery’s machine learning technology makes content searchable down to a more itemized level than legacy approaches, which tended to be limited to title, genre and a few other characteristics such as director and lead actor in the case of movies.

Content metadata creation for search is a fast-moving field that is attracting a lot of investment from the big technology players like Netflix, Google and Amazon. 3SS and Media Distillery refer to AI in the context of their partnership but their approach is strictly speaking based on machine learning, which is really advanced iterative correlation in this case applied largely to object recognition. There are however other techniques, some loosely under the banner of general AI, which are also playing roles in content metadata creation. These include OCR (Optical Character Recognition) for identifying text so that it is searchable, and at a more advanced level natural language processing for identifying context. 

Let us know what you think…

Log-in or Register for free to post comments…

You might also like...

OTT - What and Where to Monitor – Part 3

In the last two articles in this series we looked at why we need to monitor in OTT. Then, through analysing a typical OTT distribution chain, we sought to understand where the technical points of demarcation and challenges arise. In…

NAB 2019 – Technology vs Consolidation

Giving his unique view on NAB2019, Gary Olson considers and scrutinizes the big moving trends of consolidations, and casts clarity on the cloud, ATSC 3.0, and AI.

Why We Need OTT Monitoring – Part 2

In the previous article in this series, “Understanding OTT Systems”, we looked at the fundamental differences between unidirectional broadcast and OTT delivery. We investigated the complexity of OTT delivery and observed an insight into the multi-service provider silo culture. In thi…

Data Recording and Transmission: RF

In part 8 of the series “Data transmission and storage”, consultant John Watkinson looks at some of the intricacies of RF transmission.

Essential Guide: When to Virtualize IP

Moving to IP opens a whole plethora of options for broadcasters. Engineers often speak of the advantages of scalability and flexibility in IP systems. But IP systems take on many flavors, from on-prem to off-prem, private and public cloud. And…