LocateNet

  _     ___   ____    _  _____ _____   _   _ _____ _____ 
 | |   / _ \ / ___|  / \|_   _| ____| | \ | | ____|_   _|
 | |  | | | | |     / _ \ | | |  _|   |  \| |  _|   | |  
 | |__| |_| | |___ / ___ \| | | |___  | |\  | |___  | |  
 |_____\___/ \____/_/   \_\_| |_____| |_| \_|_____| |_|  
                                                         

LocateNet is a facial recognition and crowd-analysis system designed to help locate missing individuals.
The platform detects multiple faces from group photos, generates biometric embeddings, and compares them against a database to identify potential matches.

• Detects and extracts multiple faces from crowd images
• Generates 512-dimension biometric embeddings
• Similarity search using FAISS vector indexing
• Match probability scoring (high, likely, potential)
• Officer comparison gallery for visual verification
• Geographic hotspot tracking for repeated sightings

LocateNet
│
├── app.py
├── init_system.py
├── models.py
│
├── pipelines
│   └── detection_pipeline.py
│
├── services
│   ├── face_service.py
│   └── alert_service.py
│
├── vector_db
│   ├── search_service.py
│   └── storage
│
├── resources
│
└── static
    └── uploads

Clone the repository

git clone https://github.com/faizdevx/LocateNet.git
cd LocateNet

Install dependencies

pip install flask flask-sqlalchemy insightface onnxruntime opencv-python faiss-cpu numpy geopy bcrypt

Run initialization before the first launch.

python clean_system.py
python init_system.py

This creates database tables and prepares the FAISS vector index.

python app.py

Note:
use_reloader=False is used to prevent the AI model from loading twice and consuming additional memory.

1. Officer uploads a reference image
2. Face embedding (512D vector) is generated
3. Vector is stored in FAISS index
4. Database record links the vector ID to the case

1. Citizen uploads a group image
2. System detects all faces
3. Each face is cropped and stored

1. Cropped faces are converted into embeddings
2. Embeddings are searched in the FAISS index
3. Matches above threshold are recorded

Cases marked as resolved are automatically filtered from future match alerts.
This prevents unnecessary notifications once a person has been located.

• Python
• Flask
• SQLAlchemy
• InsightFace
• FAISS
• OpenCV
• NumPy
• Geopy
• Folium

See LICENSE file.

faizdevx