Sponsor Logo Detection & Forecasting Pipeline

Overview

What if you could predict how often a brand appears in a video before it even airs?

AdVisor-AI turns raw video into brand intelligence. From spotting logos in frames to forecasting their future visibility, whether you're optimizing ad placements or analyzing sponsorship impact, AdVisor-AI turns raw footage into actionable insights.

Built for analysts, marketers, and curious minds who want to see what’s coming next.

How to Run

1. Activate virtual environment:

   source venv/bin/activate

Run the pipeline: bash cd integration ./run_all.sh

🌐 How to Deploy the Website Locally

Start the Flask backend: bash cd member6_app python app.py

This will launch the backend at:

Code http://localhost:5000 Serve the frontend dashboard: bash python3 -m http.server 8080

Then open:

Code http://localhost:8080/index.html You’ll see the interactive dashboard with detection results and forecast plots.

Dependencies See requirements.txt for all required packages.

Dataset & Training Training data is located in member3_detection/dataset/ Configuration: data.yaml, readme.dataset.txt, readme.roboflow.txt Detection model: YOLOv5 (/yolov5/)

Team Members Member 1: Integration & orchestration Member 2: Frame extraction Member 3: Logo detection Member 4: KPI computation Member 5: Forecasting Member 6: Frontend

Notes

Raw video files are excluded from GitHub. Use sample frames or external links. For retraining, follow instructions in member3_detection/dataset_config/ Code

---

requirements.txt — Dependencies

torch>=2.0.0
opencv-python
pandas
matplotlib
seaborn
scikit-learn
numpy