Tracking a squat
The need for an accessible workout tool that helps improve form and keep users engaged.
What it does:
Gives real-time feedback on the user's form during workouts.
How we built it
Challenges we ran into
Integrating machine learning with computer vision isn't simple, even when trying to use a pre-trained model. Some similar technologies are even more complex or require massive technology requirements (equivalent to ~$2,400 video card) so finding the correct model and platform for our application was critical and challenging.
Accomplishments that we're proud of
This being the first hackathon for every member of the team, we are very proud of the learning we all achieved and the final product we were able to create. We learned so much about coding languages we were unfamiliar with (some members learned new languages from scratch), computer vision, machine learning, data models, and mobile UI/UX design. With our limited coding experience, we were able to research and persist through learning barriers and finish with something to show for it.
What we learned
What's next for Trackout
There is a lot of potential for TrackOut to become a huge platform to host an amazing community of users wanting to improve their workout routine. By using recurring neural networks, TrackOut will be able to provide specific and meaningful feedback to help the user achieve high levels of form and consistency with their workouts. TrackOut will also have an extensive social aspect, connecting users by allowing them to share their own workouts and help each other by providing feedback in comment sections. Finally, Trackout seeks to collaborate with major YouTube and Instagram influencers within the existing online workout space, to bring a high volume of users, and to keep them actively involed in the community.