PitchBrain is a specialized soccer tactics and analytics tool. It executes time-consuming video analysis task automatically and provides even off-the-ball analysis, which was difficult to conduct by human hand. The beta version offers three primary functions: an automatic scene detection & tagging, which uses our proprietary similarity detection algorithm; a team-style categorization & analysis; and a passing options visualization which is based on pose estimation algorithm originally developed.
Analyzing football is thought to be difficult because of the constant shifting between offense and defense, compared to other sports, so that’s why it takes so much time to analyze match videos. We, however, have developed our own algorithm that determines the degree of similarity among plays and scenes. This enables PitchBrain to automatically label & tag scenes including off-the-ball situations.
It has been possible to analyze teams’ characteristics by stats and tracking data. PitchBrain, however, is also able to recognize off-the-ball movements. This allows users to visualize team characteristics more clearly from the entire 90 minutes of a match, including off-the-ball periods of time.
With massive amount of football footages, we’ve developed a proprietary pose estimation algorithm specialized for football analysis. By this algorithm, PitchBrain can visualize passing options on each occasion with high accuracy.
For the commercial version, we will expand this function and provide new models for evaluating players, such as rating a player’s contributions to successful passes.
Hakuhodo DY Holdings Inc. and Hakuhodo DY Media Partners Inc. together formed Sports Technology Lab in November 2018. Sports Technology Lab develops innovative sports analytic solutions with the enormous sports data sets owned by Data Stadium Inc., a Hakuhodo DY Group company, and also with cutting-edge deep learning technology by Preferred Networks Inc., which formed a capitalized business partnership with Hakuhodo DY Holdings in December 2017. Sports Technology Lab is looking to expand new business ventures in the sports field both on domestic and overseas markets.
Since its establishment in 2001, Data Stadium has collected, stored, and analyzed a myriad of data from J.League, Nippon Professional Baseball, Rugby Top League, and so on, providing solutions to teams, clubs, and also players to improve their quality of plays and tactics. Data Stadium also delivers a wide range of data and data-backed entertainment content for sports fans and media outlets, thereby offering new ways to enjoy sports.
Founded in March 2014 with the aim of promoting business utilization of deep learning technology, PFN advocates Edge Heavy Computing as a way to handle the enormous amounts of data generated by devices in a distributed and collaborative manner at the edge of the network, driving innovation in three priority business areas: transportation, manufacturing and bio/healthcare. PFN develops and provides Chainer™, an open source deep learning framework. PFN promotes advanced initiatives by collaborating with world leading organizations, such as Toyota Motor Corporation, Fanuc Corporation, and the National Cancer Center of Japan.