Becoming the world’s first crypto mobile competitive environment is cool, but how do we actually bring competition to a remote environment?
Well, the key here is to remove the barriers, namely the need for a human to judge skills and the lack of a common environment and platform to enable competition.
Artificial intelligence techniques have developed massively throughout the last few years. The ability for computers to recognise sceneries, objects, actions from simple videos has become more and more widespread and achieved incredible degrees of accuracy. We have done a two-step process: we identify the region of the image that contains the player and any objects of relevance (e.g., a ball for freestyle soccer). We then draw a “box” around these objects, and within the player’s box, we run another algorithm: a pose estimator. These algorithms estimate the position of the player’s limbs and their relation to each other over time. After some fancy processing and working some of our data magic on all of this, we have a sufficiently confident estimate of every parameter needed for our scoring algorithm.
The scoring algorithm is inspired by real-life competition.
The World Freestyle Football Association defines their scoring with several parameters that judges look into, including precision, agility, innovation, and speed. Although we cannot capture all of these parameters (as some of them are somewhat subjective), we use the information we gather from the videos to score on all these different categories and provide a unified score that comprises most of what is available in the offline world.
Now it’s up to you — time to make a $MOOV and get playing. We can’t wait to have you on board.