metadata
language: en
license: other
size_categories:
- 1GB-10GB
task_categories:
- video-classification
- keypoint-detection
- image-to-video
tags:
- sports-analytics
- football
- pose-estimation
FIFA Challenge 2026 Dataset
This is the central private repository for Team 4 working on the FIFA 2026 Challenge. This dataset contains video data, skeletal tracking results, and training checkpoints.
📂 Folder Structure
| Folder Name | Description |
|---|---|
Training_Videos |
Raw video files for model training. |
WorldPose_Data |
3D coordinate data for player poses. |
Train_Data |
Processed labels and metadata for the training set. |
FIFA-Skeletal-Tracking-Light-2026 |
Light-weight skeletal tracking outputs. |
checkpoints_fifa |
Saved model weights and training states. |
DATA |
General auxiliary data and configuration files. |
compressed_videos.zip |
Archive of high-resolution video data (3.95 GB). |
🚀 Getting Started
Accessing via Python
To use this data in Google Colab, you must be a member of the fifa-challenge-team organization and use your Hugging Face Access Token.
from huggingface_hub import snapshot_download, notebook_login
# Authenticate yourself
notebook_login()
# Download the full dataset
snapshot_download(
repo_id="fifa-challenge-team/fifa-challenge",
repo_type="dataset",
local_dir="./fifa_data"
)