Futsal and movements reconized

🤝 Community Help
1

Hi Roboflow community,
We are building an AI futsal analytics platform and would like recommendations for the best architecture/workflow.
Goals:

  • player + ball detection
  • tracking
  • OCR jersey numbers
  • team recognition
  • ball possession
  • movement/action recognition
  • heatmaps and scouting metrics
    Current stack:
  • YOLO
  • ByteTrack / DeepSORT
  • EasyOCR
  • Roboflow datasets
    Main question:
    What is the best approach for futsal analytics in 2026?
    Would you recommend:
  1. Object Detection + Tracking + OCR pipeline
    OR
  2. Pose Estimation + Action Recognition pipeline?
    Also:
  • Is pose estimation worth it for futsal?
  • Should player/ball/OCR be separate models?
  • Any recommended sports datasets or pretrained models in Roboflow Universe?
    Our focus is high accuracy and near real-time full match analysis.
    Thanks!

Describe your question/issue here! (delete this when you post)

  • Project Type:
  • Operating System & Browser:
  • Project Universe Link or Workspace/Project ID:
  • Do you grant Roboflow Support permission to access your Workspace for troubleshooting? (Yes/No):
2

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.