Computer Vision Applications for Sports
by Şeymanur AYDIN
People are interested in various activities to be entertained and satisfied in order to reduce the effects of the stress and intensity of work life. Sports comes first among these activities. We can divide sports into three categories of people as “Practise sport as a profession”, “Practise sport, not as a profession” and “Not practice but follow”. According to 2019 data, the most followed sport in the world is football as 4 billion people, cricket as 2.5 billion people, and field hockey as 2 billion people. In addition, more than 200 countries participate in the Olympics, which are the world’s largest sports organization, divided into summer and winter, and more than 4 billion people follow the Olympics. The point we want to draw attention to here is that the world attaches great importance to sports and follows it closely.
While this much importance is given to sports in the world, sports organizations are expected to be satisfactory. At this point, governments, sports clubs, countries spend a lot of money for sports organizations. For example, the 2016 Rio Summer Olympics cost approximately $ 12 billion, while the 2018 South Korea Winter Olympics cost $ 10 billion. As a result of spending so much money, sports ethics should not be ignored by both the audience and the organizers, the rules are completely implemented, the referee or the evaluator can be taken wrongly, the instant follow-up and evaluation can be made in the sports that require speed and are difficult to follow, interpretations are required to be avoided.
Nowadays, there are a minimum of 5 referee errors in a match according to statistics. Lately, it has been deemed appropriate to switch from manual interpretation to digital interpretation, such as the application of video referees in football, with the wrong decisions being guiding.
Artificial intelligence technologies, which are prepared to appear in every field, also welcome us in this field. In a football match, if the foot of the footballer is close to or hits the head of the other player, the position ends with a red card. In such a case, the wrong decision may be made due to reasons such as the referee not seeing it, not being in sight. This problem can be avoided with the AI algorithm being placed on the cameras. With the “distance detection” algorithm developed by Xena Vision, instant warnings can be made and correct decisions about the position can be made.
The use of artificial intelligence supported video surveillance applications can also be given as an example of use in motorsports. For instance, Formula 1 races, which are the best known and followed most of motorsports, are in great demand. Safety becomes very important in this enjoyable but dangerous sport. There are some tracking and coordination problems arising from excessive speed and constant mobility.
In this case, it is possible to improve the process with an artificial intelligence algorithm that can analyze and observe faster than a human.. Process improvement can be achieved with “distance detection”, “accident detection” “fault detection”, “rule violation, and detection” and algorithms developed for specific problems, developed by Xena Vision.
In addition to improving sports organizations, algorithms can also be used to generate tactical tips for athletes and teams. For example, in basketball, analysis can be made such as at what angle and distance the player should jump in order to get rebound, at what angle and distance he should jump to avoid being blocked while shooting, in volleyball, at what angle to the net when it is thrown, and how far it is not touched by the net. To give an example from winter sports for the algorithm provided by Xena Vision analyzes such as posture status, body angle, and inclination, determination of the optimal posture for right and left maneuvers can also be made.
There are many cameras in sports competitions because images are broadcast from every angle. The algorithm provided by Xena Vision can work on 1000 and more cameras at the same time. Thus, the same position can be detected in more than one camera with the same algorithm and the results can be verified.
References:
https://tr.wikipedia.org/wiki/K%C4%B1%C5%9F_Olimpiyat_Oyunlar%C4%B1
https://olimpiyatkomitesi.org.tr/Detay/Olimpiyatlar/Olimpik-Spor-Dallari/49/1
https://www.bbc.com/turkce/haberler/2012/07/120723_olympics_innumbers
https://tr.wikipedia.org/wiki/K%C4%B1%C5%9F_Olimpiyat_Oyunlar%C4%B1
