SNZI is a system to detect neglected urban zones (garbage, potholes, broken roads, waterlogging, damaged infrastructure) from images and video frames using computer vision.
- Backend: Flask (Python)
- ML: YOLOv8 (Ultralytics)
- Database: PostgreSQL (planned)
- Frontend: TBD
backend/ # Flask API, DB layer
models/ # ML training + inference wrapper
frontend/ # Web UI (to be added)
dataset/ # Dataset metadata and small samples (no full data)
docs/ # Architecture and notes
scripts/ # Utility scripts
python -m venv .venv
source .venv/bin/activate # Windows: .\.venv\Scripts\activate
pip install -r requirements.txt
python backend/app.py-
GET /healthReturns service status. -
POST /upload-imageUploads an image and returns mock detections (ML integration pending).
Create .env (do not commit):
DATABASE_URL=postgresql://user:password@localhost:5432/snzid
MODEL_PATH=models/yolov8_baseline.pt
- Work on feature branches only.
- Open PRs to
dev. - Do not push directly to
devormain.
- This README is temporary and will be expanded later.
- Full datasets and model weights are not stored in Git.