How geshan/image-to-geo-location-llm works
Frontend Layer
• Streamlit UI Component (represented by streamlit_app.py)
• User interaction flow for image upload and results display
Core Processing Layer
• Gemini Integration (represented by gemini.py)
• Image Processing Pipeline
• LLM Interface
External Services
• Google Cloud Gemini API
• Cloud Run (deployment platform)
Component Organization
┌─────────────────────────────────────────────────────┐
• User Interface │
• (Streamlit Application) │
• ┬─────────────────────────┘
• ↓
┌─────────────────────────────────────────────────────┐
• Image Processing Pipeline │
• (gemini.py) │
• ┬─────────────────────────┘
• ↓
┌─────────────────────────────────────────────────────┐
• Gemini LLM API │
• (Google Cloud Platform) │
• ┘
Flow Representation
Following flows:
a. User uploads image → Streamlit UI
b. Streamlit UI → Image Processing Pipeline
c. Image Processing → Gemini LLM API
d. Gemini LLM API → Results
e. Results → Streamlit UI (display)
Container Architecture
Since the project includes Dockerfile and docker-compose.yml, show containerization:
• Docker container isolation
• How the container interfaces with external services
Deployment View
A separate section showing:
• Cloud Run deployment
• Container registry
• External access points
• Blue: Frontend components
• Green: Processing components
• Yellow: External APIs
• Gray: Infrastructure components
Integration points between
• Streamlit and the image processing pipeline
• The processing pipeline and Gemini API
• Container deployment and Cloud Run