The Premier League includes iPhones for offside decisions: a revolutionary technological change
Genius Sports has designed this new semi-automated offside technology, internally known as “Dragon.” The previous VAR system was famous for causing controversies, with “extensive delays and errors in the human process” and “concerns about the accuracy of decisions during the game.” Genius Sports has been working with the NBA for years on “optical tracking and data-driven work.”
For the EPL, Genius will start by positioning 28 iPhones in each stadium. Genius mentioned that they could add more iPhones in select stadiums throughout the year. Wired reports that mainly iPhone 14 models will be used, but Matt Fleckenstein, Genius’s product director, told them in an interview that iPhone 15 Pros will be used to carry out the task.
The iPhones will be enclosed in waterproof cases with cooling fans and connected to a power outlet. A pair of stands, each with up to four iPhones, will be placed around the stadiums. The stands will be mobile but generally left fixed during a match to record videos from a specific angle.
### How efficient is it?
According to Wired’s report, Dragon can track “between 7,000 and 10,000 data points on each player at all times.” This includes specific features such as “muscle mass, differences in skeletal structure, and even walking style,” all of which contribute to the possibility of an offside. The Verge adds that, according to Fleckenstein, other VAR systems can only use 30 or 40 points on a player. Wired states that Dragon can capture up to 200 frames per second, while current video technology reaches a maximum of 50 or 60 FPS. The system will reach a maximum of 100 FPS to “balance latency, accuracy, and costs.”
An impressive aspect of the new system is that it can also predict a possible offside. In the event of such an event, the camera’s frame rate will automatically increase to closely track the potential offside. Right after the event, it will decrease to save energy. Apparently, Dragon will also be able to input all the data it collects into a background machine learning system, internally known as “object semantic mesh.” This system will use intelligence to analyze the data and become smarter by learning from it.
