Offline AI crop assistant for smallholder tomato farmers in Kenya. Built with Google Gemma 4 for the Gemma 4 Good Hackathon.
# 🌱 FieldMind — Your Offline Field Assistant
Offline AI crop assistance for smallholder farmers.
Built with Google Gemma 4 , powered locally with Ollama , and designed for real-world low-resource environments.
---
## 📌 Overview
FieldMind is an offline AI assistant that helps smallholder tomato farmers identify crop issues using nothing more than a photo.
Built for the **Google Gemma 4 Good Hackathon**, the project focuses on **food security**, **global resilience**, and **accessible AI for underserved communities**.
Using a locally running multimodal Gemma 4 model, FieldMind analyzes crop images directly on-device and returns a simple, structured, and actionable report — even without internet access.
No cloud.
No subscriptions.
No external servers.
No data leaving the device.
---
# 🏆 Hackathon Submission
| Category | Details |
|---|---|
| **Hackathon** | Google Gemma 4 Good |
| **Organizers** | Kaggle × Google DeepMind |
| **Track** | Global Resilience / Food Security |
| **Model** | Gemma 4 E2B |
| **Inference** | Fully Local via Ollama |
| **Target Devices** | Low-resource laptops & smartphones |
### 🎯 Targeted Prizes
- 🥇 **Ollama Prize — $10,000**
Best local Gemma deployment using Ollama
- 🥇 **llama.cpp Prize — $10,000**
Best implementation on constrained hardware
- 🌍 **Global Resilience Track**
AI systems designed for difficult real-world environments
---
# 🌍 The Problem
More than **500 million smallholder farmers** worldwide lose major portions of their crops every year because diseases are detected too late.
In rural regions such as **Kirinyaga, Kenya**:
- Agricultural experts may be over **40 km away**
- Internet access is unreliable or nonexistent
- Farmers often rely on guesswork
- A single bad season can threaten an entire family's income
Mwangi, a tomato farmer, notices strange yellow spots on his leaves.
He does not know whether the issue is:
- a fungal infection,
- a pest outbreak,
- or a nutrient deficiency.
Waiting too long c …