Two categories of products used in the industry can be identified. First category of products mainly deal with a support for event tagging and drawing visualizations. One of the best examples of a first category is Once, the product which has the possibility to tag actions based on the information about action’s end result.
StatsPerform is a main representative of the second category of products supporting ball and player tracking. Except for an incredible amount of information it can provide analysts with, such platform allows analysis of opponent team without any GPS devices required. They distinguish themself from the rest of the competition by an extensive tactical analysis. Edge analysis provides the user with precise tactical insights based on the tracking data. Team shape analysis detects game moments in which opponent team makes mistakes, space can be exploited and quantifies all that with interpretable numbers for an end user. Most of the features are based on complex AI models. The PossessionValue model measures player’s impact on the probability of his team scoring in the next 10 seconds of the match. Recently, quantifying pressure players impose in a specific game moment and detailed set-piece analysis are some of the most important parts of the game on which much focus has been put.
StatsPerform uses Opta in the background to generate vast amount of match-related data. This data is collected in the blended process of combining AI computer vision models with human annotators who thorougly correct system’s inefficiencies. Spiideo uses AI on a level above, to obtain high-quality panomaric footages without human intervention required. There are many other video analysis software solutions like Hudl, Dartfish, Longomatch, Nacsport etc., no offend to others not mentioned. Except for StatsPerform, more-less all systems are mainly visualization tools wih only a basic capabilities of generating data. However, they really excel in creating powerful visualization, tagging capabilities and its immense number of dashboards that can be created.

This blog post gives a basic introduction to industry-level products useful for the football analysis. In the next blog post, I will try to explain in more details how object detection is useful for such tasks. Until the next time 🫡.