Compute vision and neural networks: MEGOGO to launch new promo technologies

16 June 2016

The Company promises a breakthrough in advertising business MEGOGO online cinema team started testing new technologies built upon computer vision and neural networks, meant for bringing media advertising to a new level in the nearest future. We speak about image identification technology in a streaming video. The Company believes it’s possible to change the very approach to advertising by means of innovations. They should simultaneously decrease advertising load on the video platform users and increase advertising efficiency in several times.

New evolution advertising formats to be launched by MEGOGO are as follows: profound targeting, cut-in and advertising targeting in TV-streams, connection with e-commerce platforms, and extended content on the “second screens” and in players, direct communication with viewers, built upon Big Data etc.

MEGOGO is going to present technologies and launch first advertising campaigns in a revolutionary format this autumn.

According to Ivan Shestakov, MEGOGO Marketing Director, CIS advertising market currently undergoes growth stage, and there is a need for changing the very paradigm of an advertising message. ‘Although modern advertising Internet formats contain unique features, they are still the offsets of old good offline advertising, merely adjusted to the possibilities of browsers and apps. Today we, being one of the largest advertising market players, are brave enough to state we can gain much a different momentum to the advertising industry development’.

MEGOGO has started talking seriously about revolution recently with emergence of the first test results of a new image identification technology in a streaming video. They claim that have put some share of their film library through algorithms using educable artificial neural network and got hundreds of thousands of marks (tags), which with an extremely high degree of probability identify items, scenes, people profiles and their actions within a video sequence of a film or show. ‘It’s hard to explain how we became inspired upon receipt of the first results. Of course, new algorithms do not represent sound artificial brain like in fantasy and we cannot pass Turing test yet, but it actually can identify Turing in newsreel fragments and even what he is wearing on and is doing on the video. And that, to say the least of it, is unbelievable!’, – says Ivan Shestakov, MEGOGO Marketing Director.

Thus, as soon as new algorithms process the entire video service library and get with TV streams, MEGOGO team will get enough possibilities to offer an advertiser a profoundly targeted advertising on its platform, which, besides, will be shown not only to the targeted audience, but will appear on a screen depending on what is going on there. Currently the algorithm identifies thousands of images and now, for example, can clearly define where a car of a certain class, brand and model appears in a film, a person of a certain sex and age, let us say, speaking on the phone, or just a cup of steaming coffee. All that remains is to show an advertising in the right place at the right time.

A targeted advertisement message is just the beginning. MEGOGO is going to do the following steps:

  • Context connection to e-commerce platforms so users could immediately order goods and services they see on the video;
  • Extended information in TV content as far as the platform analyzes video on-the-fly, even on air on TV and can present additional info on the second screen – for example, of a smartphone or a tablet while a user watches a TV show.

Moreover, based on these technologies in the nearest future there will be systems of complemented reality along with VR-technologies, being worked upon by MEGOGO developers over the last year. During this period MEGOGO has released a number of applications to watch 3D video and 3600 video for virtual reality glasses.

New advertisement technologies being worked upon by MEGOGO imply deep video content analysis, comparing received data with the array of information on behavior and interest of each separate user. ‘If we know our audience perfectly, identify each separate pair of eyes and understand needs of their owner, then we can show content of that very interest. Such awareness of the user and content allows defining the most suitable time and place for perception of specific advertisement messages’, – convinced Ivan Shestakov. Beyond any doubts, advertising budget effectiveness directly depends on the relevance of an advertising product and content, correctly targeted on a separate user.

MEGOGO convinced that success of an advertising model of the project depends on three factors: audience analysis and profiling, deep content analysis and its consumption principles, as well as readiness of the advertiser to accept new advertising formats. And if everything is clear with the advertisers’ readiness – they comparatively quickly accept new possibilities if they see them effective, then in terms of technology everything is more complicated.

It’s not a problem anymore for technology companies to collect all possible users’ data. And it’s a little bit harder to analyze huge data arrays. However, today MEGOGO extensively applies mathematical statistical models when acquiring content, promoting one or another library positions, establishing individual communication channels with each user. ‘A natural development step of a video platform is to apply deep analysis Big Data in advertising business. As far as this is being done in recommendations and content acquisition, so why not to use the same methods to enhance advertising model, at the same time making it better for the advertiser and user?’