metadata

license: mit

HockeyOrient Dataset

🔗 This dataset is part of the HockeyOrient ecosystem.

💻 Check out the corresponding Hugging Face Space for a live demo: https://huggingface.co/spaces/SimulaMet-HOST/HockeyOrient
🏒 The trained model for this dataset is available here: https://huggingface.co/SimulaMet-HOST/HockeyOrient

Overview

The HockeyOrient dataset consists of 9,700 manually annotated frames of ice hockey players from the 2023–2024 Swedish Hockey League (SHL) seasons. Each cropped image is categorized into one of eight player orientations based on the orientation of their head.

Dataset Features

Classes: 8 orientations – Top, Top-Right, Right, Bottom-Right, Bottom, Bottom-Left, Left, Top-Left.
Annotations: High-quality manual annotations.
Sources: Extracted frames from SHL games.
Formats: Images in JPEG format organized into folders for each class.

Structure


HockeyOrient/
├── Bottom_Left/
│   ├── <Unique_ID>.jpeg
├── Bottom/
│   ├── <Unique_ID>.jpeg
├── Bottom_Right/
│   ├── <Unique_ID>.jpeg
├── Left/
│   ├── <Unique_ID>.jpeg
├── Top_Left/
│   ├── <Unique_ID>.jpeg
├── Top/
│   ├── <Unique_ID>.jpeg
├── Top_Right/
│   ├── <Unique_ID>.jpeg
├── Right/
│   ├── <Unique_ID>.jpeg

📩 For any questions regarding this project, or to discuss potential collaboration and joint research opportunities, please contact:

Mehdi Houshmand: [email protected]
Cise Midoglu: [email protected]
Pål Halvorsen: [email protected]