13.04.2017
In the traditional competition of start-ups, which took place in the framework of the annual conference Venture Day Minsk on April 12, the new Belarusian project Fansy won. It teaches artificial intelligence to find interesting moments in game video streams. As a prize, the team received tickets for the TechCrunch Disrupt conference in Berlin.

In the traditional competition of start-ups, which took place in the framework of the annual conference Venture Day Minsk on April 12, the new Belarusian project Fansy won. It teaches artificial intelligence to find interesting moments in game video streams. As a prize, the team received tickets for the TechCrunch Disrupt conference in Berlin.

This year, 40 teams from Belarus, Russia, Ukraine, Bulgaria, Lithuania and Israel applied for the start-up competition. In the finale, the 11 best startups performed before the pool of international investors.

A feature of the current final was the participation of several well-known and bright teams. Among them were OneSoil, who developed a smart system for increasing yields, the online travel designer Eightydays.me and the “green” charging HandEnergy, for which the 19-year-old Gomel man raised $ 70 thousand on Kickstarter.

However, the winner became a project, about which few people have heard so far, called Fansy. The second place went to OneSoil, the third place took Exponenta, which gives the possibility to predict the vitality of unpublished texts and finalize them.

Fansy team develops an algorithm based on artificial intelligence, which compile the best moments from the long streams of video games. None of the competitor startups does this automatically and in real time, right at the stream time, claim the creators.

“We have developed an artificial intelligence that watches video games and makes highlights automatically,” said Fansy founder Konstantin Keller. “We are the first in the world, this nobody has ever done.”

According to him, today more than 2 000 000 streamers are represented on platforms like Twitch, Beam, Azubu and YouTube Gaming. To expand the audience on other channels –  YouTube, Facebook and Twitter –  they need to adapt the broadcasts by making the video shorter. To do this, you will have to look through the entire stream and find interesting moments. Video editors take for this work about $ 100 per video. Fansy is an electronic video editor that uses cloud infrastructure and computer vision. “Infrastructure costs us very cheap, about $ 0.02 per hour of video processing of high quality,” the founder of the project underlines.

Today 6 people work on the product. SRT Sergey Anishchenko has more than 10 years of experience in the field of computer vision, image analysis and machine learning. He received a PhD degree from the University of Middlesex (Great Britain), works at the Research Institute of Neurocybernetics named afterA.B. Kogan in Rostov (Russia) and in parallel is engaged in the start-up. CEO Konstantin Keller spent three years in one of the largest gaming developers in the CIS.

An early prototype of Fansy was created during Minsk AI Hackathon in December 2016. Then for four months the team has been preparing a version of the product, which is currently undergoing beta testing, and the first users are really excited.

The startup has already been accepted into the famous Finnish accelerator, the training will start in May. While the development is carried out at the expense of their own money (the founders invested $ 10,000), but in the summer the team plans to attract the first investment.

The winner of the last year's contest was the Belarusian project KUKU.io, which developed the service for SMM to manage accounts in social networks.

Source: dev.by