Hugging Face's logo

Spaces:
trainmatricx
/
README
Running

App Files Community
README / README.md
Asifikbalprotik's picture
Asifikbalprotik
Add Train Matricx organization card
064ba28 verified
preview code
|
Raw History Blame Contribute Delete
2.25 kB
metadata 
title: README
emoji: ⚽
colorFrom: blue
colorTo: indigo
sdk: static
pinned: false

Train Matricx

Human-verified training data for sports computer vision.

Train Matricx is a managed sports data annotation team. Sports AI teams send match video and a labeling schema; annotators who know the sport produce the labels, and every delivery goes through a 3-layer QA pipeline before it ships.

We are a service team, not a software platform. We work inside your schema and your tools, or deliver in the format your pipeline already ingests.

What we annotate

  • Player and ball tracking with persistent player IDs through occlusion, camera cuts and substitutions
  • Event tagging to your own taxonomy (passes, shots, set pieces, serves, rallies and sport-specific sub-events)
  • Skeletal keypoints and pose for movement and biomechanics models
  • Court and pitch registration for mapping broadcast pixels to field coordinates
  • Segmentation and bounding boxes for detection models

Delivered as COCO, YOLO, JSON, CSV or your internal format. 20+ sports covered, including football (soccer), basketball, cricket, tennis, ice hockey and golf.

How we check quality

  1. Primary annotator applies the labels to your class definitions
  2. Specialist reviewer cross-checks complex events and verifies player identity across occlusions
  3. Senior auditor confirms the delivery matches your schema exactly

We report quality as first-pass acceptance: the share of deliveries your team accepts without rework.

Work with us