Analyzing retail store customer traffic using cutting-edge machine learning.
EyeSee is our graduation project for our B.Sc. in Computer Science at Bar-Ilan University. The project leverages advancements in computer vision and machine learning to analyze retail store customer traffic and provide actionable insights.
We customized and trained the Ultralytics YOLO model to perform three major tasks:
- 👥 Person Detection and Tracking: Identifies and follows individuals within the store.
- 📅 Age Classification: Categorizes detected individuals into age groups.
- ⚥ Gender Classification: Determines the gender of detected individuals.
These capabilities enable EyeSee to analyze customer traffic patterns from video footage, offering detailed insights into store activity.
EyeSee provides the following main features:
- 📊 Detailed Reports: Generates comprehensive reports based on analyzed footage.
- 📈 Dashboard: Displays weekly and yearly trends in customer traffic.
- 🔥 Heatmaps: Visualizes customer activity within the store for better spatial insights.
Check out our video on YouTube!
$PROJECT_ROOT (EyeSee)
├── Client
│ # Client-side code
├── Server
│ # Server-side code
├── VisionModel
# AI model files (code, weights, etc)
- Frontend: React with Material UI
- Backend: Node.js with Express
- Database: MongoDB (Atlas)
- Machine Learning: Python with TensorFlow/PyTorch and Ultralytics YOLO
- Media Management: Cloudinary