How coaches and managers are using AI to build winning teams and track player fitness

From tweaking the composition of a team to preparing for medical exigencies during a multi-sport event, AI fixes are getting better and are being more widely deployed the world over. (Image via Pexels/Pixabay)

Technology has become increasingly important in all areas of the sports industry, including scouting, training, performance analysis, fan engagement, broadcasting, and management.

Over the past two decades, coaches have increasingly turned to data science to boost their teams' overall performance and making split-second judgments. For example, referees in football increasingly use Video Assistant Technology to help them make more accurate calls on crucial plays like penalties, free kicks, and red cards.

The speed at which computers can process data has also contributed to sports fans' insatiable desire for new strategies and tools. Now with advances in deep learning, the sporting experience is set to undergo even more radical transformations. Let us explore some of them:

With clubs in major and minor leagues alike able to use tracking systems to better understand their players' activities, the global player tracking market is predicted to develop at a CAGR (compound annual growth rate) of 24.9 percent over the next five years.

Predictive analysis is one way AI is utilized in sports to improve performance and health. For example, wearables that monitor stress and fatigue are helping athletes stay healthy.

But this is only the beginning. Technology based on artificial intelligence can aid teams in developing strategies and tactics and maximizing their capabilities.

AI has allowed for a far more nuanced analysis of player performance. Coaches can exploit the opponent's weaknesses by adapting their tactics and plan based on the day's data and visualizations depicting their team's strengths and shortcomings. For example, using human posture estimation, the effectiveness of swimmers has been analyzed using single stationary cameras above and below the water filters. The current quantitative evaluation method, which requires human annotations of body components in each video frame, is time-consuming and will be phased out in favour of this system.

The Special Olympics employed SAS AI technology to analyze the flood of incoming data and anticipate when health problems would arise, allowing them to position medical staff and supplies strategically throughout the venue. In addition, each participant was given a smartwatch that transmitted their health and location data via the Internet of Things (IoT) to a centralized dashboard so doctors could keep tabs on them.

The Chief Intelligence Officer of the 2019 Special Olympics World Games in Abu Dhabi, Yousef Alhammadi, stated in a case study published by SAS that "1,529 medical occurrences happened during the weeklong event, some of which were significant episodes including seizures and asthma attacks... Sometimes, life or death was decided by how quickly we get an answer, and analytics significantly contributed to that effort."

Similarly, the use of AI in weight training has demonstrated some encouraging benefits in the research. Apps like FoodVisor already use deep learning to recognize over 1,200 food items, calculate an estimated serving size, and generate a nutritional analysis in seconds.

Now that we have the computer vision method of human pose estimation, we can teach computers to recognize human positions in real-time. Other areas where this tech has been used include online yoga and pilates, where keypoint skeleton models are utilized to identify human joints and offer the user guidance on exercising correctly.

This information was derived from measurements taken by 15 relatively inexperienced individuals as they performed three to five sets of 10-12 repetitions on a leg press machine. Competent coaches evaluated the trainees' progress and found promising results in both performance and prediction. In addition, it demonstrated the viability and effectiveness of using AI methods to automatically assess performance on weight training equipment and offer timely suggestions to athletes.

Sports bettors have been trying to forecast the outcomes of future events by analyzing massive amounts of data for years. For example, tennis statistics have been studied to better forecast game outcomes, including first- and second-serve percentages, aces, backhand winners, and more.

Unfortunately, humans can't keep up with the amount of data generated by AI-driven football algorithm predictions. They also can't make enough correct predictions to amass enough wealth through betting. The constraints inherent in their humanity ensure they will never collect a fortune in the millions. Even AI can't correctly forecast the outcome of every match. But, if it uses an algorithm to make predictions, it can get much closer than a person.

Researchers, for instance, tested several setups of the pos-N-M model on 4,000 manually labelled frames and found that it attained an accuracy of 85.5 percent. Although a few things were working against them, like the fact that the ball was small and hard to see or was sometimes hidden by another player.

It is clear how important data and machine learning have become in sports. However, despite more technology in sports today, sports can still tell us how people live. For example, technology makes it easier for athletes to live healthier lives and gives competitive teams an edge.