maldima11/maize_yield_project

A Hybrid AI framework for predicting maize yields in data-scarce environments using Generative AI, Transfer Learning, and multi-channel dissemination (USSD, Mobile, Web) for smallholder farmers in Zimbabwe. # maize_yield_project A Hybrid AI framework for predicting maize yields in data-scarce environments using Generative AI, Transfer Learning, and multi-channel dissemination (USSD, Mobile, Web) for smallholder farmers in Zimbabwe. Phase 7: Inclusive Dissemination and Designed Interaction Architecture This phase focuses on the "last mile" delivery of the AI model's insights. Because smallholder farmers in Sub-Saharan Africa experience varying levels of digital literacy and internet access, this framework deploys predictions across multiple inclusive channels: a Web Dashboard, a Zero-Data USSD/SMS system, Voice AI (IVR), and an offline-capable Mobile App. 📂 Folder Structure To prevent dependency conflicts between the Python AI backend and the Dart/Flutter frontend, the repository is structured as follows: maize_yield_project/ │ ├── backend_api/ # Python Flask Server (AI Engine, Mock API, & USSD Logic) │ ├── venv/ # Python virtual environment │ ├── app.py # Main Flask application and API routes │ ├── ussd_handler.py # Africa's Talking USSD/SMS logic │ └── requirements.txt # Dependencies (Flask, requests, tensorflow, etc.) │ ├── web_dashboard/ # Interactive Web Interface │ ├── app.py # Main dashboard application (Streamlit) │ └── requirements.txt # Dependencies specific to the web UI │ ├── mobile_app/ # Flutter Application (for Extension Officers) │ ├── lib/ │ │ ├── main.dart # Flutter app entry point │ │ ├── services/ # API connection logic │ │ └── database/ # SQLite local caching logic │ └── pubspec.yaml # Flutter dependencies │ └── notebooks/ # Jupyter/Colab Notebooks for Model Training ├── 01_data_extraction.ipynb └── 02_model_training.ipynb 🛠️ Prerequisites & Local Setup (macOS) To run these interfaces locally on your Mac, ensure you have the follow …