floorballstudio
FI|EN
OPEN SOURCE TOOL

MediaPipe

Cross-platform ML framework for live/streaming media, including pose and hand landmark tasks.

Fast client/on-device option for quick pose and hand analysis before heavier server processing.

Technology

C++ / Python / JavaScript-adjacent Tasks APIs, on-device ML

License

Apache-2.0

Permissive; keep license/notice requirements. Suitable for on-device and web-adjacent pipelines.

GitHub ↗

Implementation / installation

Use MediaPipe Tasks packages on the target platform; keep video local when privacy is a priority.

Coaching use

  • Quick skeleton preview during recording.
  • Hand landmark estimation for grip-width analysis.
  • Client-side quality checks before video upload.
Floorball Studio

Try the related prototype

This project is evaluated as a building block. The Floorball Studio prototype keeps the user experience sport-specific.

Open prototype →