TeamSportz uses Google platform TensorFlow to detect player form and movement
5 min read

TeamSportz uses Google platform TensorFlow to detect player form and movement

Tech
Aug 23
/
5 min read

It’s no secret that AI is revolutionising the sport and fitness industry through the many wearables and apps that are popping up in the space, enabling people to train and perform (from wherever) like never before. While the industry is without a doubt starting to get crowded, avid basketball player and established software developer, Francisco Baptista, noticed a problem that needed solving and saw a gap in the market that needed filling.


The problem is that the amateur sports market was largely underserved with technology that would enable players, coaches, and teams to reach their full potential. Looking to break down the barriers, Francisco began developing something that would be the answer to this problem and launched TeamSportz, an innovative sports AI platform enabling players, teams and sport academies to improve their performance and understanding of the game whilst enhancing health and wellbeing.


TeamSportz is the first AI sports app that is multi-sport, built for the amateur market, secure and safe for kids, and easily accessible and affordable. The level of innovation and accuracy of the app would not be possible without the push from the giants like Facebook, Apple and Google, who are making artificial intelligence more accessible to everyone.




AI making it all possible

Diving into the tech a little bit more, TeamSportz uses the Google TensorFlow platform to deploy its models to both web and mobile. What this model does is it detects the ball and the player using the pose detection model, MoveNet, which in combination with TeamSportz sport-performance algorithms, result in accurate performance statistics.



When we say accurate performance statistics, we really mean it. Using TensorFlow’s MoveNet model, TeamSportz is able to detect focal points and exact angles of the player’s body, as well as quick movements made by the player. The combination of AI detection and interactions make it so no movement will be missed when players are completing an exercise or drill (and they can’t cheat either)!


In speaking with TeamSportz CEO, Francisco Baptista, he shares, “In our mission to democratize sports performance data, we create simple solutions for coaches to configure exercises and even create new exercises with just a few clicks and share with their players in a matter of minutes. Coaches now have the ability to easily track ball control and performance exercises remotely and therefore work more efficiently and closely with their players to help them improve their game”.


https://www.youtube.com/watch?v=7VlCpOirK9Q

What’s it like under the hood?

The TensorFlow MoveNet model provides the pose detection of the body in the following format (for all of you non-techies reading this, don’t worry if this looks like gibberish to you):


TeamSportz then creates a unique set of angles for every adjacent part of the body and using a state-machine algorithm, they’ve created a simple Analysis Editor that enables coaches to customise what part(s) of the body should be tracked to then generate accurate performance stats.



Francisco is set to achieve even more by working closely with the TensorFlow team, as it “will enable [him] to try and test some great features coming out next, such as the multi-pose-estimation, 3D estimation and more.”


If you’re a coach, we encourage you to sign up for TeamSportz Coaches Corner and see how much easier it makes your job to coach your team. Stay tuned for more updates and innovations with the TeamSportz app. This is just the beginning.

TeamSportz uses Google platform TensorFlow to detect player form and movement
5 min read

TeamSportz uses Google platform TensorFlow to detect player form and movement

Tech
Aug 23
/
5 min read

It’s no secret that AI is revolutionising the sport and fitness industry through the many wearables and apps that are popping up in the space, enabling people to train and perform (from wherever) like never before. While the industry is without a doubt starting to get crowded, avid basketball player and established software developer, Francisco Baptista, noticed a problem that needed solving and saw a gap in the market that needed filling.


The problem is that the amateur sports market was largely underserved with technology that would enable players, coaches, and teams to reach their full potential. Looking to break down the barriers, Francisco began developing something that would be the answer to this problem and launched TeamSportz, an innovative sports AI platform enabling players, teams and sport academies to improve their performance and understanding of the game whilst enhancing health and wellbeing.


TeamSportz is the first AI sports app that is multi-sport, built for the amateur market, secure and safe for kids, and easily accessible and affordable. The level of innovation and accuracy of the app would not be possible without the push from the giants like Facebook, Apple and Google, who are making artificial intelligence more accessible to everyone.




AI making it all possible

Diving into the tech a little bit more, TeamSportz uses the Google TensorFlow platform to deploy its models to both web and mobile. What this model does is it detects the ball and the player using the pose detection model, MoveNet, which in combination with TeamSportz sport-performance algorithms, result in accurate performance statistics.



When we say accurate performance statistics, we really mean it. Using TensorFlow’s MoveNet model, TeamSportz is able to detect focal points and exact angles of the player’s body, as well as quick movements made by the player. The combination of AI detection and interactions make it so no movement will be missed when players are completing an exercise or drill (and they can’t cheat either)!


In speaking with TeamSportz CEO, Francisco Baptista, he shares, “In our mission to democratize sports performance data, we create simple solutions for coaches to configure exercises and even create new exercises with just a few clicks and share with their players in a matter of minutes. Coaches now have the ability to easily track ball control and performance exercises remotely and therefore work more efficiently and closely with their players to help them improve their game”.


https://www.youtube.com/watch?v=7VlCpOirK9Q

What’s it like under the hood?

The TensorFlow MoveNet model provides the pose detection of the body in the following format (for all of you non-techies reading this, don’t worry if this looks like gibberish to you):


TeamSportz then creates a unique set of angles for every adjacent part of the body and using a state-machine algorithm, they’ve created a simple Analysis Editor that enables coaches to customise what part(s) of the body should be tracked to then generate accurate performance stats.



Francisco is set to achieve even more by working closely with the TensorFlow team, as it “will enable [him] to try and test some great features coming out next, such as the multi-pose-estimation, 3D estimation and more.”


If you’re a coach, we encourage you to sign up for TeamSportz Coaches Corner and see how much easier it makes your job to coach your team. Stay tuned for more updates and innovations with the TeamSportz app. This is just the beginning.

Francisco Baptista
Architect

Quisque imperdiet tellus morbi sodales nullam enim amet. Enim gravida cursus facilisis sodales id mattis vestibulum gravida magna.