News & Events Sports Technology

Sports Tech Shoutouts – May 2021

For our May edition of Shoutouts, it's all about computer vision technology and what 3 different organisations are doing in that space. We also highlight a few podcasts episodes that are sort of in-line with our theme. They touch on technologies that track performance (including computer vision) and the practical applications and outcomes of them.

Hey hey and welcome to the May edition of our Sports Tech Shoutouts. For those who has had a crazy or hectic past month, know that you are not alone and do reach out to someone if you can. Mental health decline is a growing issue especially after the lengthened period of trying to overcome the pandemic through lockdowns and restrictions. Also, for those that read, I highly recommend the book Shoe Dog by Phil Knight. It’s his personal memoir of how he started Nike. There isn’t any of those “10 steps to become something great” or “tips and tricks for making stuff”, it is simply a candid recollection of that part of his life and the journey he took to put Nike together. Still I think it is gold, and I am only half way through the book. There is this part of the story where he was overwhelmed by the challenges of running the shoe start-up and starting a family and he basically told himself: “Life is growth. You grow or you die.” I find that so relatable. The journey of life can be challenging, even tough, on the other hand it also makes a lot of sense that we need to go through the challenges to grow. I reckon we are meant to learn and grow in life. We may not do that at the same pace as others but we are meant to grow.

Pose Estimation Using Computer Vision

But back to technologies, I have had the privilege of having some interesting conversations with enterprising sports tech folks in the last month who are developing motion tracking solutions using cameras. That combined with what we observed over the last year or so, I can confidently say that the application of computer vision for sports and fitness tracking is on the rise. When pose estimation technologies using (3D) cameras became available to consumers (think Primesense or Microsoft Kinect) back in early 2010s, there were mixed reactions to it. Some thought the tracking could replace trainers or coaches. Some considered the possibility to use it for biomechanical analysis. But it never quite happened for a few reasons. Accuracy was one of it. Fast forward to today, although pose estimation algorithms have improved heaps, Motion Capture systems with markers are still the ‘gold standard’ for biomechanics. Also, we see that coaches make use of the tech to assist them in making better quantified decisions but the delivery of the individualised coaching is still 100% the coach and no tech can quite replace that. Instead what we are seeing is an integration of technology together with guidance from experts (coaches or biomechanists or physical therapists) for specific areas.

In this edition of Shoutouts, we feature some of the enterprises working in this space. One thing they have in common is their goal to be non-hardware specific, and for their solution to be deployable on as many smartphones/tablets/laptops as possible. You should definitely read further if you are either:

  • An athlete or coach in Baseball, Softball, Golf or Cricket exploring performance analysis solutions,
  • A developer interested in developing solutions with the latest pose detection/estimation algorithms,
  • A startup or established firm in health and fitness wanting to integrate an existing motion tracking solution, or
  • A sports professional wanting to learn more about the application of tech in sports

Sports Trace Launches Mobile App For Analysing Swings And More

Sports Trace: Baseball, Softball and Golf analysis

SportsTrace is a startup that has developed a performance analysis solution for the individual athlete and they have recently launched their mobile app on the App Store and Google Play. In one of the co-founder’s own words, they call this solution “Moneyball in motion” and it is smart automated video analysis that helps athletes and coaches succeed. This is done through taking video footage of the athlete’s motion, uploading it to the cloud where the images get analysed and the SportsTrace platform provides actionable insights and personalised recommendations to the athlete/coach.

Although SportsTrace officially launched in early 2020, the idea of tracking athletes with a camera was born much earlier when one of the co-founders, Max Montrey, saw an opportunity with using Kinect to track joint movements in sports. Having played baseball and also coached baseball, Max tinkered with the idea of using the Kinect camera to analyse baseball swings. It remained a hobby slash side project until he got approached by accelerators to commercialise it. That coupled with seeing the potential demand for such a solution pushed it to the next level.

SportsTrace App Screens

Recognising the limitations of the Kinect and seeing the potential democratisation through using smart phone cameras, Max and co-founder, Alex Gardner focused on further developing computer vision solutions on 2D images. Besides developing the tech, a key part of the solution is compiling/curating a library of knowledge around the sports; partnering with experts including biomechanists, coaches and sports researchers. So when an athlete uploads a video, the SportsTrace (cloud) platform will analyse the motion, breakdown the kinematic sequence, then provide individualised recommendations that will help them improve optimally. They currently provide solutions for athletes and coaches in Baseball, Softball and Golf. They have also started exploring adding other sports with an obvious parallel being Cricket.

So if you are in any of the above sports, and are looking (or open) to adopting an assistive tech solution to take your game to the next level, download their app or reach out to Sports Trace on their website. Also, read more about how their solution can be helpful here: link or check out their demo video below:


Google Research Launches Next-Gen Pose Detection With MoveNet And TensorFlow

Google has recently published a real-time pose detector with MoveNet and TensorFlow.js, which can run on a web browser on a mobile device or laptop with a camera. Sure, pose estimation or detection over a web browser is not new. The big key differentiator here is that it is ultra fast (or ultra real-time), and it is much more accurate than traditional detectors especially when it comes to difficult poses. The gif below shows it quite clearly.

Comparison of a traditional detector (top) vs MoveNet (bottom) on difficult poses. (source: blog.tensorflow.org)

Other interesting key points to note in this pose detection system/engine are:

  • It detects 17 key points/joints on a body
  • There are two variants – “Lightning” is for latency-critical applications while “Thunder” is for higher accuracy requirements.
  • Both variants can run at higher than 30 frames per second (FPS) on most modern devices
  • There are no dependencies to install.

As it says on the TensorFlow blog: “The result is a model that can deliver accurate keypoints across a wide variety of poses, environments, and hardware setups.

There are huge potential applications in sports, health and fitness with this setup. TensorFlow has already teamed up with IncludeHealth to power their interactive web application for remote care. Even with the early release of MoveNet, IncludeHealth was able to quickly see the benefits of this new pose detector over previous detectors – speed and accuracy. Besides remote allied health care, this is equally valuable in remote coaching or personal fitness training. On top of that, in activities such as dancing or martial arts (e.g. karate) where executing precise movements are evaluated by the naked eyes of the judges or examiners, this might provide the real-time objective measure that could help with the final decision.

Going forward, the TensorFlow team are looking to continue optimising their backends to accelerate model execution across all supported devices. They are also looking to extend Lightning and Thunder models to the multi-person domain, so that developers can support applications with multiple people in the camera field-of-view. In the meantime, for the developers who would like to have a go at it, they have provided a tutorial here. There are also more details about this release here: link or check out this video about TensorFlow below:


VAY Developed A Customisable Plug & Play Motion Analysis Platform (For Fitness & Health Providers)

At this point in 2021, most of us would be quite familiar with doing many things remotely – working remotely, learning remotely, doing telehealth and doing virtual workouts. Allied health or fitness providers who were typically face to face and had minimal digital presence or a digital platform would have started something digital during 2020. It may be virtual live group workouts, streaming workout content, or one-on-one video call consultations for personal training or rehabilitation.

Now with motion tracking technology made possible simply using a web camera or mobile device camera (+ computer vision algorithms), it is a great opportunity for health and fitness providers to add-on such a technology to provide additional value in their services. No, they do not need to hire a computer vision expert to build a solution from the ground up because there are ready-made solutions out there that just requires integration.

VAY is a startup based in Switzerland and they have developed a motion analysis platform that sports, fitness and health providers can integrate (or plug in) to their existing websites or platform. They started off creating a fitness tracking/coaching app and launching it on Google Play and App Store. But they pivoted shortly after to come off the consumer market and instead provide their solution to businesses because they believed they would have a bigger impact that way. For businesses that want to adopt their solution, they provide full integration support so that the businesses can hit the ground running. A couple of other advantages of Vay’s platform are 1) it can be run locally or on the cloud, and 2) they have a continually growing exercise library and allowing new movements to be added depending on their client’s needs.

So if you are a fitness provider, allied health provider, or equipment manufacturer and you considering integrating a camera tracking solution, reach out to them on their website. You can also read more about their motion analysis solution here: link; or have a look at their demo below:


One Last Thing – Podcasts

One last thing before I end – I wanted to highlight a few sports tech related podcasts that have some interesting content and related to the topics we discussed in this edition.

  1. Bullpen Presents – In Conversation With – Episode 006 – SportsTrace (link)
  2. FUTRSPRT – Episode 182 – Sukemasa Kabayama, CEO Uplift Labs (link)
  3. One track Mind – Episode 07 – What Technology Do We Wish Existed? (link)

And that is our sports tech shoutouts for May 2021. If you would like more information about any of the above, or if something sparked an idea and you would like to chat about it, feel free to reach out or leave a comment below. If you enjoy our content, please do share it using the links below or subscribe to our blog here: link. As always, thanks for reading!

Thoughts? Leave a comment here.

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: