Skip to main content

HockeyOrient

A Dataset for Ice Hockey Player Orientation Classification

Domain
Sports
Task
Pose estimationClassification
PublicationGitHub repoLast updated Jun 2026

HockeyOrient: A Dataset for Ice Hockey Player Orientation Classification

Dataset Model Demo Paper

🏒 Overview

The HockeyOrient dataset consists of 9,700 manually annotated frames of ice hockey players extracted from the 2023–2024 Swedish Hockey League (SHL) seasons. Each cropped image is categorized into one of eight player orientations, based on the direction of their head relative to the camera. This dataset is designed for player direction classification, which can enhance tracking models, tactical analysis, and AI-driven coaching tools.

📂 Dataset Details

  • Classes: 8 orientations – Top, Top-Right, Right, Bottom-Right, Bottom, Bottom-Left, Left, Top-Left.
  • Annotations: High-quality manual annotations.
  • Source: SHL game footage.
  • Format: Images stored in JPEG format, categorized into separate folders for each class.

📥 Access the Dataset

The dataset is hosted on Hugging Face:

🔗 Dataset Path: HockeyOrient on Hugging Face

🎯 Pretrained Model

We provide a trained classification model based on SqueezeNet, available on Hugging Face.

🔗 Trained Model: HockeyOrient Model

🚀 Live Demo

Explore the HockeyOrient classification demo on Hugging Face Spaces.

🔗 HockeyOrient Demo: Hugging Face Space

🤝 Citation

If you use this dataset or model in your research, please cite:

@inproceedings{HockeyOrient,
    title = {{HockeyOrient: A Dataset for Ice Hockey Player Orientation Classification}},
    author = {Houshmand Sarkhoosh, Mehdi and Gautam, Sushant and Midoglu, Cise and Sabet, Saeed Shafiee and Kupka, Tomas and Halvorsen, Pål},
    publisher = {Association for Computing Machinery},
    booktitle = {Proceedings of the 16th ACM Multimedia Systems Conference},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3712676.3718342},
    isbn = {9798400714672},
    location = {Stellenbosch, South Africa},
    year = {2025},
    doi={10.1145/3712676.3718342}
}

📩 Contact

For inquiries, collaborations, or questions:

⚡ HockeyOrient is part of ongoing AI-driven sports analytics research.