[ad_1]
“In volleyball, we now use state-of-the-art cameras to track not only athletes, but also football,” says Alain Zobrist, chief of Omega Timing. “That’s why we use camera technology and creative ingenuity.”
The Omega Timing R&D department has 180 engineers, and development strategies began with the installation of a machine gun, according to Zobrist, in 2012. The goal was to reach a point where, in a series of 500-plus games that take place each year, Omega can deliver details of what they are doing in the running game. This data should also take less than one-tenth of a second to test, process, and send to the event so that the information matches what viewers are seeing on the screen.
With beach volleyball, this meant taking the skill of navigating and driving and teaching AI to identify several types of shooting – from smash to blocks to spikes and variations – and across colors, as well as the way to enter the ball, and then combine this data with the information gained from to gyroscope sensors in player clothing. These driving sensors allowed the system to know the movements of the runners, as well as the length of the jump, the speed, and much more. Once prepared, all of this is fed to advertisers for use in response or on the screen.
According to Zobrist, one of the most difficult subjects for AI to learn was to accurately follow the ball when the cameras could no longer detect it. “Sometimes, it is covered by a part of the athlete’s body. Sometimes it goes off the TV, ”he says. “As a result, the problem was to track the ball when you lost it. To have the app predict where the ball is going, and then, when it reappears, re-read the difference between the lost ball and return it, and fill [missing] data then proceed automatically. That was a very difficult subject. ”
It is this pursuit of football that is essential for AI to realize what is going on in the game. “Once you follow the ball, you know where it was and when it changed direction. And with the combination of speed sensors, algorithms detect shooting, “says Zobrist.” Whether there was a block or a smash. You know which team he was and which player he was. Hence the combination of all these technologies makes us more accurate in comparing the amount of data. ”
Omega Timing says its volleyball form on the beach is 99% accurate, thanks to sensors and multiple cameras that move at 250 frames per minute. Toby Breckon, a professor of computer science and photography at Durham University, however, is interested in seeing if this represents the game — and, in particular, whether the system is being deceived by racial and gender differences.
“The experience is amazing. And you need a lot of knowledge to teach AI in a variety of areas,” says Breckon. “But one of those things is correct. How often do they err on the side of caution? How often does he lose his temper? Also if it works the same for all races and genders. Is that the 99% accuracy, say, a group of women in the USA and Is 99% accurate in the Ghanaian women’s group? ”
Zobrist is confident, explaining that while it may be easy to call Google or IBM to provide the necessary AI expertise, this was not the way for Omega. “The most important thing, whether it’s a game, or a time-lapse game, is that we can’t tell the difference between the description and the outcome,” he says. “In order to protect the integrity of the results, we cannot rely on any other company. We need to have expertise to be able to explain what happened and how the athletes got there.”
Speaking of future changes and follow-ups, Zobrist is strong, but he says the Paris Games in 2024 will be important. “You will see new features. Obviously, it stays around time, scoring, and around motion sensors with a built-in interface. Also in Los Angeles in 2028. We have some interesting activities there that we have just started. ”
Many Great Stories
[ad_2]
Source link
