Home - cnnsyhnx/CrowdInsight GitHub Wiki
CrowdInsight
CrowdInsight is a modular, AI-powered people analytics library that detects and analyzes individuals, groups, emotions, and demographics in video streams. Built for scale and privacy, it's suitable for public institutions, governments, retail, and smart-city applications.
π βFrom busy malls to border surveillance drones, understand who is passing through β without invading privacy.β
π Wiki Index
- π Getting Started
- 𧱠Architecture
- π€ Outputs
- ποΈ Use Cases
- π¬ Model Details
- βοΈ Configuration
- π§ͺ Testing
- π€ Contributing
- π License
π§ What Can It Do?
- Detects people, pets, objects, accessories
- Analyzes gender, age group, emotion
- Outputs JSON for analytics or dashboards
- Designed to be privacy-friendly
- Works with both live and offline footage
𧩠Key Technologies
- YOLOv8 (Ultralytics) for object detection
- DeepFace for emotion/gender/age
- OpenCV for image preprocessing
- Torch & NumPy backends
- JSON export for downstream use
ποΈ Built For
- Government surveillance (e.g., borders, transit zones)
- Smart cities and crowd management
- Businesses analyzing visitor patterns
- Hotels, malls, public events