How ecocity-the-try-people/ecocity-traffic-prediction works
Project Type & Purpose
• This is a single-service, Python-based image-processing component of a larger “Ecocity Traffic Prediction” system.
• It fetches traffic images from Firebase, runs YOLOv8 inference to detect vehicles, computes a congestion level, and writes results back to Firebase.
Key Components & Boundaries
a. External Image Source
• Could be IoT cameras or an upstream service that pushes image URLs into Firebase Realtime Database or Cloud Storage.
b. Firebase (Google Cloud)
• Authentication & Credentials (via firebase_admin + JSON key)
• Realtime Database (stores image URLs and congestion results)
• (Optionally) Cloud Storage for actual image blobs
c. Processing Service (main.py)
• Environment loader (dotenv)
• Firebase client (firebase_admin)
• Image fetcher (requests, io, PIL)
• YOLOv8 inference engine (ultralytics, model file yolov8n.pt)
• Congestion calculator (simple vehicle count → level)
• Database updater (writes back to Firebase)
d. Developer Machine / Scheduler
• Cron job or manual trigger that runs python main.py
Architectural Patterns & Principles
• Single responsibility: The service only handles image analysis & DB updates.
• Stateless processing: Each run fetches data, computes, writes results without local persistence.
• Dependency injection via environment variables: Credentials and DB URLs configured in .env
• Model as a resource: yolov8n.pt loaded at startup or per invocation