This project involves annotating football match videos to track player movements using bounding boxes across video frames. Using Label Studio, each player (Liverpool, Arsenal, Referee) was manually tracked through sequences, creating a rich labeled dataset ideal for training object detection and tracking models in sports analytics.
The primary goal was to:
- Identify and annotate players from different teams.
- Track their positions frame by frame.
- Prepare structured data usable for AI model training (e.g., YOLO, DeepSORT, etc.).
- Tool Used: Label Studio
- Video Frames: Manually annotated sequences over 114 frames.
- Labels:
Player_Liverpool,Player_Arsenal,Referee. - Annotation Format: JSON export from Label Studio.
Each annotation captures:
- Frame-specific bounding boxes.
- Movement sequences.
- Label identity (team/referee).
✅ Hands-on experience with video annotation pipelines.
✅ Understanding of object detection labeling standards.
✅ Practical application of Label Studio annotation workflows.
✅ Prepared data suitable for computer vision model training.
Through this project, I have learned:
- How to structure annotations for computer vision tasks.
- The importance of precise bounding boxes in model accuracy.
- Best practices for managing large-scale annotation tasks.
- How annotated data feeds into AI-driven sports analytics solutions.
To visualize my work, here's a sample screenshot from my annotation session:
If you are interested in leveraging my skills for your AI or data annotation projects, feel free to reach out via LinkedIn: