SPAI-Marketing (SPAI Marketing Sports Promotion through AI Driven Marketing) is an AI-powered platform designed to evaluate sponsor logo visibility in professional football match footage. Built with modularity and scalability in mind, this system combines state-of-the-art computer vision techniques with real-time data analytics to deliver meaningful insights for sports marketing stakeholders.
Check out a demo video of SPAI-Marketing in action:
👉 Watch the Demo
This project integrates several key components:
-
🎯 YOLOv8-based Logo Detection
Trained to identify and locate sponsor logos on football kits with high precision using annotated match frames. -
📊 Sponsorship KPI Engine
Custom logic to quantify sponsor exposure based on screen position, visibility level, and logo size throughout the match. -
🎥 Video Processing Pipeline
Built with OpenCV and DeepSORT for object tracking across frames, enabling logo visibility aggregation over time. -
🌐 Interactive Dashboard
A Plotly Dash-based frontend that displays game-specific visibility metrics, dynamic graphs, and aggregated reports for each sponsor. -
🗃️ Database Integration
Uses Supabase (PostgreSQL) to manage uploaded games, sponsor metadata, and analysis results across the platform.
- Logo detection powered by YOLOv8
- Real-time logo tracking and visibility scoring
- Sponsor exposure metrics (duration, size, screen position)
- Match-specific dashboards
- Admin interface for uploading game videos and metadata
- Seamless backend/frontend communication via FastAPI
- Scalable pipeline adaptable to other sports or media assets
- Python (YOLOv8, OpenCV, FastAPI)
- Plotly Dash (Frontend)
- Supabase (PostgreSQL DB)
- Google Colab (Model Training)
- Roboflow (Annotation & Dataset Management)
SPAI-Marketing/
│
├── Notebooks/ # Google Colab notebooks for training, inference, testing
├── SPAI_admin/ # Admin dashboard (Dash + FastAPI integration)
├── SPAI_client/ # Client dashboard to visualize sponsor metrics
├── .gitignore # Git configuration
├── README.md # Project documentation
├── requirements.txt # Python dependencies
This project is proprietary and closed-source.
All rights reserved © 2025 Ghayda Tebessi.
Unauthorized use, copying, or distribution is strictly prohibited.