How nano-a/medicinal-plant-classifier works
Project type and purpose
• Type: Full-stack web application with an embedded Machine Learning component (RandomForest)
• Purpose: Classify plants as medicinal or non-medicinal based on user-entered botanical features via a Flask web interface; also generate evaluation visuals (ROC, confusion matrix, feature importance).
File structure
• Top-level
• app.py: Flask entry point (“main”)
• requirements.txt, .gitignore, devbook.md, Tache.txt
• data/: CSV datasets (plants.csv, plants_extended.csv)
• model/: ML logic and pipeline (plant_classifier.py)
• controller/: Flask routes (routes.py) – implements MVC “Controller”
• view/: Optional Jinja result template (results.html)
• templates/: Main Jinja template (index.html) – “View”
• static/: CSS, images, visualization outputs
• generate_images.py: Standalone script to regenerate static evaluation graphics
• Architectural pattern: Monolithic Flask app following MVC conventions, single process, file-based data store, no external DB or microservice.
Extract insights from README
• Tech stack: Python 3.10, Flask, pandas, scikit-learn, matplotlib, seaborn, Tailwind-inspired CSS
• Workflow:
1. User fills web form (index.html)
2. Controller routes.py processes POST → calls plant_classifier
3. plant_classifier performs preprocessing, model inference, evaluation metrics
4. Controller returns results (json or through view template) + visual assets (static/images)
• Additional: generate_images.py can be run offline to produce ROC, confusion, feature importance plots stored in static/images.
• Client / Browser (UI)
• Flask App Server
• app.py (Bootstrap & Configuration)
• Controller (routes.py)
• Model Layer (plant_classifier.py)
• Visualization Generator (generate_images.py)
• Data Store (CSV files in data/)
• Static Assets Server (CSS, JS, images under static/)
1. Client → HTTP(S) GET → Flask Server → Serve index.html + style.css
2. Client submits form → HTTP POST → Controller
3. Controller reads data (plants.csv), invokes Model Layer → model.predict()
4. Model Layer applies preprocessing, returns prediction + metrics
5. Controller assembles results, embeds dynamic visualizations or links to static plots
6. Flask Server → HTTP response → Client (renders results.html or JSON)
c. Architectural patterns & design principles
• MVC separation: Templates (View), routes.py (Controller), plant_classifier.py (Model)
• Single-process, modular code organization
• File-based data store for simplicity; can swap for a database if needed
• Offline batch visualization script decoupled from request cycle
d. Key technologies/frameworks
• Flask (web framework & routing)
• pandas, scikit-learn (data processing & ML)
• matplotlib, seaborn (charts)
• Jinja2 (templating)
• Tailwind-inspired CSS (static/css/style.css)
Elements
• Clear, concise labels for each component (e.g., “plant_classifier.py”)
• Rectangle for code modules