Swimming is the most popular Olympic sport and is being practiced by millions of people. Currently swimming is being transformed by SwiMMinD, a startup bringing artificial intelligence to swimming lanes.
The concept of SwiMMinD sprang in the autumn of 2017 by Lida Tsonkova (Learn to Swim expert with experience in America, Australia, Spain and Bulgaria,) and Valentin Milkov (US Open Champion and Manager of a World Cup in Swimming). Soon after that, the team joined Borislav Popov, an Artificial Intelligence Expert with over 17 years of experience of managing tech teams. The multiple Olympic and World Championship medalist in swimming Thomas Lurz also joined the team. Mentor of this team is Yong Kyu Lew, Professor at Hankuk University, specializing in technology, innovations and partnership management. This world-class team aims to use streaming data to apply machine learning, deep learning and artificial intelligence to capture and analyze the collected data, making way for innovations in learn to swim and remote swim races.
Today our lives are intertwined with digital media and it is difficult to define how long a person should be in front of a computer or mobile phone screen. Childhood obesity and diabetes continue to grow in alarming rates. Research reveals that children consume between 3 to 6 hours of media in a typical day. This includes using computers, cell phones and tablets for games, music and reading. Unfortunately, children tend to avoid exercising!
In this respect, SwiMMinD promises to connect the digital world with the world of swimming. It focuses on the beginner swimmers and shall help them achieve their goals and empower their full aquatic potential. In other words, SwiMMinD aims to populate the swimming pools because it promises to help millions of people learn to swim or race globally in an easy, fun way!
For an example, SwiMMinD will help any kid check whether he has learned to make a star in the water. Once the star is confirmed by SwiMMinD that kid will receive an interactive digital certificate based on well-established Learn to Swim standards from Australia, America and England. That certificate, for instance “Gold Fish” will come to live and tell that boy what needs to do to pass the next level.
An example for the Race Video Analysis Module is when a kid already knows how to swim. His father video records his 50 meters Freestyle in the local pool in San Diego, America and enters him into the global race against a kid in Melbourne, Australia.
Traditional video analysis companies focus on the elite level swimmers because the tools being used are time consuming. In other words, it is difficult for coaches to provide video analysis to all of their swimmers because it takes hours with those drawing tools to make the analysis. SwiMMinD thinks that faster analysis is better. Therefore, SwiMMinD shall offer AUTOMATIC video analysis!
In summary, SwiMMinD shall automatically deliver meaningful FEEDBACK and interactive digital AWARDS to beginner swimmers interested to build confidence in the water and shall provide them with video based competitions to RACE with swimmers around the world. To achieve this SwiMMinD shall evaluate swimming videos by using machine learning algorithms to calculate whether a swimming skill was executed properly and whether and how fast a swimmer swam an event. This will augment coaches tremendously and will bring more joy to swimmers and parents.
As a consequence of all of the above, in January 2018 SwiMMinD won “Most Business Value” Smarter Award from the International Olympic Committee and Alibaba Cloud. Part of the Award was exhibiting SwiMMinD at the Olympic games in Pyeongchang, South Korea.
Machine learning and deep learning
SwiMMinD models on body movements in the water using machine learning algorithms, which by nature improve on performing a task as they gain more experience. Thus, SwiMMinD shall automatically assign a value to each action, such as a star in the water. A perfectly executed star, for example, has a high value, but a star with bended arms or legs would have a lower value.
Valentin Milkov, the Co-Founder of SwiMMinD says that the artificial intelligence and machine learning will play an important role in the swimming analytics in general. “Existing mathematical models develop existing knowledge and insights in swimming, while AI will make it possible to discover new connections that would not be obvious for coaches, swimmers or parents. Thus, SwiMMinD will not only help people Learn to Swim based on well-established international standards but alsoSwiMMinD will offer the optimum swimming execution of a stroke”
SwiMMinD shall distinguish body parts and movements by accurately compiling images from videos. The event stream processing shall enable real-time image recognition using deep learning models. “By combining our deep learning models, we could train our models in memory in the cloud” says the CTO of SwiMMinD, Bobi Popov.