CombineAI

Use your physical performance to create a professional sports profile using AI.

Comment

Inspiration

Young athletes often struggle to showcase their incredible talents and abilities to scouts and their peers, making it harder for them to excel in their chosen sport. We wanted to build an AI platform which simplifies the process and allows them to immediately produce a professional profile, present themselves with confidence and secure a place in the team they've always wanted to join.

What it does

The platform is built on three pillars.

  • No. 1: The platform accesses the webcam or takes an uploaded video, and the user chooses an activity to complete/measure
  • No. 2: The user's movements are tracked. The AI uses kinematic and biomechanical feature analysis to decide the user's scores for each activity and shows them live on screen
  • No. 3: The user has the option to save their score and generate a report, which uses their score obtained in each activity. The users' data is saved to an SQL database, and it is then fetched and displayed on the scout dashboard. Scouts can then use the filters to choose specific athletes.

How we built it

We built the website with Python (Flask to turn it into a functioning website) and HTML (using the built-in 'style' and 'script' tags to make the frontend work with the backend). We built the biomechanical feature extraction using OpenCV (to access the user's camera and turn it into data we could work on) and MediaPipe (to use that data and extract the angles and distances of each joint). We used Claude AI (Code) to help us with some parts of the programs.

Challenges we ran into

  • Getting the camera feed to have as low a delay as possible
  • Getting the data extraction to be as accurate as possible
  • Integrating multiple components (e.g. backend, frontend, Flask) as this required quite a lot of debugging and testing

Accomplishments that we're proud of

  • We integrated real-time AI analysis using multiple tools and made them work smoothly together
  • The scoring system we made actually made personalised recommendations instead of just generic profiling
  • Most importantly, we made an end-to-end product that has efficient ergonomics and generates an actual report based on data. This aligns with our aim to help young, talented athletes showcase their skills .

What we learned

  • How to apply Euclidean geometry and Vector mathematics in the real world (in this case how to use it to calculate angles and exact movements)
  • How to translate a live video feed into a result through a complex backend

What's next for CombineAI

In the future, we wish to work with larger organisations so that our product can be used in the real-world sports industry. We will also add additional features such as more activities to provide a better UI and UX. We hope that our product will be used by real athletes and scouts so that they can succeed in their chosen sport.

Share this project:

Updates

Submission history