Gazprom-Media Holding: Video Tagging System

The main articles are:

• Biometric identification technologies
• Speech Recognition (Technology, Market)
• Speech technology: On the path from recognition to understanding
• Video analytics (terms, applications, technologies)

2023: Service Launch

Gazprom-Media Holding"" April 5, 2023 announced the launch recognitions video content of the Video Tagging System automatic and tagging service. It was developed by the TV channel "!" and Friday the company "" GPM Date- a single asset management center big data "-Gazprom Media Holding."

The Video Tagging System is based on computer vision technologies and neural networks. The service helps to recognize faces, speech (speech recognition is carried out using Tinkoff Voice Kit technologies), inscriptions and the number of persons in the frame, movements of people, perform detection of animals and objects; classify cinematic plans, read the beginning of screensavers and credits, and identify the prevailing stage palette.

All this allows you to speed up the search for the necessary personnel for the production of promotional materials and the launch of contextual video advertising. The product automates the search for suitable scenes, moderation, subtitling and content analytics of video content. The system is capable of automatically marking the most relevant moments for displaying commercials (midrales), as well as marking logical pauses in multi-part content.

The first users of the service were the TV channels "Friday!" And "Saturday!," As well. As online movie theater PREMIER a result of using the VTS system time , the selection of video materials for creating promo videos was reduced by more than 20 times, and the subtitling time of content was reduced by 4 times. Using VTS reduces work when working with content and allows users to redistribute resources from routine operations to creative tasks.

TV channel "Friday!" actively uses the technology to create promos.

{{quote 'author = said Maxim Rusakov, head of Internet projects "Fridays!" | A few years ago, colleagues from GPM Date showed us the possibilities of neural networks, and the idea arose to use these miracles to solve the operational business problems of Friday!. At a minimum, automate and speed up the work of promo producers. As a result of joint efforts, the VTS service appeared. The UMA video platform has become a partner for distributing and storing video, as well as demonstrating the capabilities of contextual video advertising. As of April 2023, VTS searches the video archive from 3000 hours of content. With this service, promo producers spend a couple of minutes searching for video clips, and earlier it took weeks to do such work,}}

In March 2023, the Saturday! TV channel was the first to test the technology in contextual video advertising. An advertisement for the confectionery show with Klava Coca "Hello, Sweet" was integrated inside the show "Confectioner" on "Friday!." Advertising automatically appeared on the air after the words "sweet" and "try." So during the viewing, the audience "Pyatnitsa!" saw the relevant promo of another confectionery project.

{{quote 'author = noted Svetlana Lapunova, Managing Director of GPM Date. | We are constantly expanding the portfolio of our digital products for Gazprom-Media Holding and advertisers. Solutions based on computer vision and artificial intelligence technologies have high potential in optimizing work with media content and during the configuration of native advertising communications for online video. The demand for such developments will grow not only within Gazprom-Media Holding, but also from external customers,}}

{{quote 'author = noted Sergey Stamboltsyan, PREMIER CEO. | PREMIER already uses the subtitling technology offered by the VTS system. This helps to optimize the work of preparing promotional materials, as well as adapt content for hearing-impaired viewers. PREMIER has great ambitions for the development of service technologies, so we plan to actively implement the work of the new system and test its capabilities as part of working with content,}}