SoccerMetrics

• 

  Good shot

• 

  Bad shot

Inspiration

I wanted to enhance soccer training by combining computer vision and analytics.

What it does

SoccerMetrics analyzes a player's kicking form, computes angles at impact, and offers personalized tips.

How I built it

I used Python (Flask, OpenCV, Mediapipe) for the backend and React/TypeScript for the frontend, with YOLO for object detection and the GHUM 3D CNN based model for pose landmarks.

Challenges I ran into

Handling video processing efficiently and synchronizing pose data with the ball's position were our biggest hurdles.

Accomplishments that I'm proud of

I successfully combined pose estimations, ball tracking, and a coaching AI to deliver actionable insights.

What I learned

I deepened my understanding of computer vision, machine learning workflows, and real-time data processing for sports analytics.

What's next for SoccerMetrics

I plan to expand features for other soccer techniques, add advanced AI models, and refine the user interface for broader adoption.

Built With

• flask
• mediapipe
• numpy
• openai
• opencv
• pandas
• react
• typescript
• ultralytics
• yolov8