AI : A game changer for the sports industry
FIFA is an emotion to those who follow football like a religion. When controversial goals happen, like the infamous one of Maradona’s ‘Hand of God’, the chaos is inevitable. Enter artificial intelligence technology, which not only helps avoid such controversies but also, enables referee to make accurate offside calls with the help of computer vision and sensors tracking player & football motion.
Let's take a look at how Artificial intelligence can be used with regards to football and how different types of annotation techniques are required to train such AI models.
FIFA this year, has announced the use of AI-powered cameras & have planned to implement semi-automated system which will involve placing sensor into the football itself. This sensor will relay the footballs position on field 500 times per second. Along with the sensor, there will 12 computer vision camera in different locations across the stadium which will use 29 points on the body of the players to track their motions. When the player is in offside position, this information will trigger an alert to the official pointing out that they are closer to the other team’s goal post than their second last opponent & are receiving the ball. This alert will enable the official in the control room to validate & notify the referees on the pitch.
The above process will take a few seconds and would not have any effect on the timing of the game. It will also, enable offside calls, to be made with faster & greater accuracy.?
Monitoring the movements of individual players to gauge their performance is an important aspect for teams to perform better in tournaments like FIFA, Champions League & La Liga. With computer vision based AI models, trainers & coaches can observe & analyse the sessions & provide feedback on performance to the players. This can be done in concert with other sports analytics programs with ML algorithm enabled cameras.
This method is reliant on accurately annotated video training data & can be applied to scouting methodologies, to spot talent before competition. Skeletal annotation can track movements of body parts though each video frame & interpret a player’s form & movements from these datasets.
领英推荐
Machine learning algorithms also enable coaches to analyse the tactics at a greater level of detailing. Artificial intelligence models trained with annotated videos of team movements & interactions can be used to identify the patterns of competing team. Bounding box, a method of data labelling, can be used to process video data & allow annotators to quickly locate individual players over the large area of football field.
Ever wondered, how FIFA on your PS is such a realistic game? With player moving naturally & with fluidity?
There was a time when game developing companies had to rely on professional athletes in motion-capture suits to perform various actions such as dodging, dribbling & passing the ball. The players were required to do anything that they would do on a football field. This process was extremely time-consuming, costly, tedious & lets, face it, not practical enough.?
With the increase in the development of motion capture technology, the file sizes have increased as well. It has become almost impossible to collect every possible motion combination.
This is where the motion sequence segmentation comes into play. Companies can now record actual football matches & annotate them. Human motion data adjusts multiple actions in sequence & temporal segmentation of human sequence into plausible motion primitives is of extreme importance to complex motion annotation. DesiCrew annotators identify the independently moving objects in the video & separate them from background motion. If there is a need for fine details such as facial expressions, our annotators go for landmark annotation. This is where key points are placed along with person’s face to track things like eye movements & facial expressions.
DesiCrew provides high quality annotations at the leading edge of sports & fitness AI. Reach us on [email protected] to know more.