README.md

metadata

title: HockeyAI
emoji: 🏒
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.11.0
app_file: app.py
pinned: false
license: mit

🏒 HockeyAI: A Multi-Class Ice Hockey Dataset for Object Detection

🔗 This interactive demo is powered by the HockeyAI model and dataset.

• 📂 Download the dataset used in this project: https://huggingface.co/datasets/SimulaMet-HOST/HockeyAI
• 🤖 View details of the trained model: https://huggingface.co/SimulaMet-HOST/HockeyAI

HockeyAI is a specialized dataset and object detection system designed for ice hockey analysis. Built on the YOLOv8 architecture, this project provides accurate detection of key hockey game elements including players, officials, and game-specific features. The project includes both a comprehensive dataset and benchmark implementations using YOLOv8.

🎯 Dataset Classes

Our dataset includes seven key classes essential for hockey game analysis:

• Center Ice (centerIce)
• Face-off Circles (faceoff)
• Goals (goal)
• Goaltenders (goaltender)
• Players (player)
• Pucks (puck)
• Referees (referee)

📊 Model Specifications

• Architecture: YOLOv8 Medium
• Framework: Ultralytics YOLOv8

🔧 Usage Guide

1. Upload any ice hockey game frame
2. The model will detect and classify:
   • Game elements (center ice, face-off circles, goals)
   • Personnel (players, goaltenders, referees)
   • Equipment (pucks)
3. View results with bounding boxes and confidence scores

💻 Technical Implementation

• Backend: Python 3.9+
• Interface: Gradio 5.11.0
• Deep Learning Framework: PyTorch
• Hardware Optimization: GPU-accelerated inference

📩 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]

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference