Since 2018, Swimming Australia has worked with Amazon Web Services (AWS) to develop technology that is the envy of other global swimming powerhouses. The technology, coupled with the drive and determination or our swimmers and the knowledge and experience of some of the best coaches in the world, has led Australia to punch well above its weight.

Swimming Australia is using computer vision, machine learning and generative AI to track swimmers and analyse their results.  The technology has changed how coaches and athletes approach competition and training. By using data-driven insights, it provides valuable information for coaches to make better informed decisions about their athletes and tailor training programs for individuals to be the best they can be.

It has used the technology at major meets around the world – including the past two Olympic Games – with the results speaking for themselves, including 18 medals in the pool at last year’s Paris Olympics.

In the past, Swimming Australia has only used the technology at high-level race meets, however, in its push to continue to innovate and stay ahead of the rest of the world, the technology will now also be used in the training pool – another world first.

The initiative will help coaches monitor and adjust training sessions in real time, even where multiple swimmers are sharing lanes and performing different drills.

The announcement was made at AWS Re:invent in Las Vegas in December. Swimming Australia has termed it Training Insights and the technology will eventually be rolled out to all 14 high performance centres around the country.

One of the main drivers of the technology is Jess Corones, the General Manager of Performance Support and Olympic Campaign at Swimming Australia.

Jess told ITWire in Las Vegas that: “Training Insights is going to be ground-breaking in what we can do. We will be able to analyse the whole squad over the whole training session with the results in real time. It will enable the athlete and coach to track their training in detail we have never had before.

“Our coaches are some of the most knowledgeable in the world. Their depth of understanding of the sport, their knowledge and experience and intuition are world class. We learn so much from listening to them so being able to provide them with something like Training Insights to help them in their coaching. We think we are going to understand the sport in a way that we have never done in the past. It is really exciting what the future holds.”

In recent years Swimming Australia has also upped the ante with its race data. This has become possible by upgrading its single pan camera – which is attached to the ceiling of the swimming pool - to a larger, higher resolution 4K camera which identifies and tracks every athlete in every lane – including non-Australians.

“Every country is doing their own thing but they don’t have the sophisticated systems that we do,” says Jess. “And I don’t think any other country could analyse all eight swimmers in a race into the depth of data that we are able to and with the speed. We can turn around a 100m race, all eight swimmers, fully analysed in 20 minutes and get that to the coaches.”

Swimming Australia’s state-of-the-art custom-built system is known as Lane 4 – a web portal that centralises performance data, making it easily accessible across devices. An integral part of that is Sparta - a race analysis system that uses machine learning and generative AI to track swimmers' performance.

Vision from the camera is combined with this race analysis technology which tracks the entire field and provides crucial real-time metrics such as stroke distance, stroke rate, speed as well as analysis of the more technical components such as starts and tumble turns.

For the swimmer, the Sparta data is overlayed with the vision of the event so the athlete, coach and performance analyst can analyse it.

“We want the athlete to look at the vision with the data. The data will tell them everything about their race from number of strokes, length of the strokes, the number of kicks and so on. It will tell us if there were any personal best performances or areas where they have not been able to execute the race plan. We then want them to go “oh yes, I remember that part of the race. This is what I did” because we want them to be able to repeat it the good parts and recognise and improve specific parts.”

There is no doubt Australia is leading the world when it comes to this type of technology. It has partnered with AWS since 2018 to develop these ground-breaking tools which are the envy or other top swimming nations as well as many of Australia’s other elite sports.

The partnership with AWS came about in 2018 when the then Chairman and President of Swimming Australia, John Bertrand, secured a sponsorship deal for services support. Not long before Jess had attended a Rugby Australia presentation which showcased how they were bringing all their data from around the country into one central location.

“It made me think what could Swimming Australia do if we could harness our data sets from all these different points that we had.”

Up until then Swimming Australia’s performance analysts had to manually analyse the swimmer’s performances by forwarding a video forward one frame at a time.

“When we started working with AWS – it was a big learning experience for us. I remember the first meeting and we asked AWS “what do you do and what can you do for us”. They ran some workshops – what they called The Art of the Possible. They taught us to think outside the box and think about what would you do if you had access to this information.”

Jess said the big breakthrough for Swimming Australia came at the 2021 Tokyo Olympics where it rolled out the first application they built with AWS called Relay App which uses AI and machine learning to help predict and provide scenario suggestions when selecting relay teams.

“In Tokyo we won a medal in all seven relay events. I don’t know whether that had been done before. I don’t think that would have been possible without the data insights that we had and the speed at which we were able to get those data insights to the coaches to support them with their decision making in who they thought should swim and in what order.

“This allowed them more time to prepare and make sure we had the best strategy going into every race.

“It was quite satisfying getting feedback from other teams – “we can’t believe you had this tool at the Olympics and we didn’t know about it”. The Americans came up to me and said “so you knew what we were going to do before we knew what we were going to do”. I said “yep”.

Lane 4 uses AWS’ SageMaker with data going into QuickSight to analyse and visualise the vast amounts of data generated and provides actionable insights for coaches and athletes. It is where the coaches go to for all the data visualisation - looking at the trajectories of their athletes, world standards, stroke efficiency and race pacing – all in one place, a place to be curious.

“Ultimately the coaches are trying to improve human performance every day and with the technology we have they are able to dive really deep into the data and get answers to any questions they have to enhance the coaching process,” says Jess.

Jess said she was not worried about other nations copying what Swimming Australia is doing. “Swimming is a global race both in and out of the pool. We would be naive to think that other countries are not trying to replicate what we are doing and that is part of the fun of it as we try to stay ahead so we need to keep innovating. With AWS, the cadence with which we are able to work – we are able to move a lot faster with our innovation, prototyping and getting those products and tools into the hands of the coaches, athletes and performance staff a lot quicker.”

Jess said the technology was particularly useful for relays. “Each relay has its own head coach so, for example, Dean Boxall is the coach of the women’s 4 x 200m freestyle team. Dean and I will meet most days to discuss the relay. Dean might say he is thinking of swimming Mollie O’Callaghan first, Jamie Perkins second and so on and Ariarne Titmus last. I put that order into the system and run the analytics and it will say this is what we think the outcome of that race will be – it will tell us where we will place.

“We also shuffle the order to see if we get a different outcome. It has all the teams so it will say this is the US team and this is the order the computer thinks they will swim.

“Sometimes you don’t go with the fastest stat. This is where it’s a combination of intuition and experience of the coach versus the data. The computer says if everyone swims at their best this will be the outcome but we have to consider the psychology of it as well. A computer will never be able to takeover fully because it doesn’t know, for example, when a swimmer comes in and tells the coach they have been sick all night. Or if they are walking around pool deck really nervous and fidgety – the coach can see that – and that is where they use their experience and intuition.”

Jess said the technology works just as well for all strokes and no matter the length of the race.

“The technology works great in events such as the mixed relay because there are so many different scenarios and it becomes very tactical. For us to do all that long hand it would take me three months. Now because of machine learning I can just put all the data in and hit a button and have all the different scenarios played out within 30 seconds.

“At the Tokyo Olympics the coach came up with a scenario but then the computer told us to do something else. We spoke about it and the result was a combination between what the computer was telling us and what the coach thought from their experience.”

The issue for the team was around Emma McKeon, who was going to swim the butterfly leg but two events before the relay Emma was swimming in a 50m freestyle race. The computer did not know that Emma was going to swim the other event so the coach decided to keep Emma on the same stroke – freestyle. “We ended up swimming Matt Temple in the butterfly and Emma in the freestyle and they got a medal. It was amazing!”

Jess said there had been no push back from the coaches about the technology. “It actually helps them and gives them confidence in the decisions they are making. They have all these ideas and all this pressure on them to make the right decision. If they have the access to data that confirms their thinking it can give the coach confidence that they have the data to back up the decision that they are making.”

Our final question to Jess was around how successful the team would have been without the technology. “We will never know but I think the data is definitely there to enhance and support the process. I think it guides them and I think as we move forward we are trying to find better ways to use the data. I think in the future it will become a more embedded process for coaches and it will become such an invaluable tool for them that if you don’t have the data you are going to be guessing at some point. The data is there as a co-pilot for the coach.”