As the Australian Swim Team prepares for the Paris Olympics in July, they are set to leverage cutting-edge computer vision technology. This advancement aims to enhance their training regimen with predictive insights, building on the success they achieved at the 2021 Tokyo Olympics.
Harnessing Data Post-Tokyo Success
After securing nine gold medals in Tokyo, Swimming Australia, the sport's national governing body, acknowledged the significant impact of their data analytics investments. Jess Corones, the general manager of performance support and Olympic campaign, highlighted that since 2021, AWS has played a crucial role in linking their data lake, Atlantis, to a wider array of data sources and developing innovative predictive tools for coaches.
"The Tokyo Olympics success underscored the transformative power of data on performance, leading to further investments," Corones remarked. The preparation for each elite swimmer, known as 'Dolphins', was informed by Atlantis's predictions on diet, training schedules, injury management, and race strategies.
For the first time, both historical and real-time performance data were utilized to optimize relay team compositions and orders, resulting in the Dolphins clinching medals in six out of seven relay events at the Tokyo Olympics.
Advancing Computer Vision in Training
Since 2019, Swimming Australia has employed its "Swim Performance and Race Tactical Analysis" (Sparta 2) system to monitor athletes, primarily during competitions. The current focus is on integrating this technology into training sessions. The objective is to automatically capture swimmers' techniques and performances during training and incorporate this data into Atlantis's performance insight models.
"We're working with AWS Professional Services to develop 'Training Insights,' a system that employs machine learning and image recognition to track athletes in training pools," Corones explained. This initiative aims to provide comparable data metrics between training and competition settings.
Tracking metrics such as distance-per-kick, breath count, and turn times is more straightforward in competitions due to consistent variables. However, training environments pose challenges with multiple swimmers and varied routines, which the new system seeks to overcome.
Lane Four: Revolutionizing Data-Driven Coaching
In 2018, Swimming Australia partnered with AWS to transform athlete performance data into actionable insights for coaches. This data encompasses statistics from the International Swimming Federation Association (FINA), athlete management systems, wearables, and Sparta 2.
The AWS Professional Services team utilized the AWS Serverless Data Lake Framework to integrate these data sources, creating comprehensive data lake environments. This integration has facilitated the development of more individualized training programs tailored to each athlete and coach.
In early 2023, Swimming Australia and AWS launched an online portal, "Lane Four," providing coaches with access to swimmer data and predictive insights via an AWS QuickSight dashboard. "Lane Four" serves as a comprehensive resource for coaches, accessible on their iPads, offering information on athletes' strength, conditioning, and nutrition.
The portal also supports natural language queries, enabling coaches to quickly access specific data, such as the top three fastest turns in the women's 100-meter freestyle since 2003. This functionality is crucial during training sessions, providing immediate benchmark data to set goals for athletes.