Inspiration
Small-scale Kenyan farmers lose 30-40% of their crops to preventable diseases. I wanted to democratize agricultural expertise using AI technology that's already in farmers' pockets.
What it does
Agricheck analyzes photos of diseased plants and provides instant diagnoses with organic treatment plans using locally available materials in Kenya.
How it's built
Frontend: Streamlit for a mobile-friendly web interface AI Engine: Google Gemini 2.0 Flash for multimodal image analysis Localization: Custom prompts optimized for Kenyan crops and organic remedies
Challenges
Gemini model compatibility issues requiring multiple API versions Balancing AI accuracy with practical, actionable advice for farmers Secure API key management for deployment
Accomplishments
Created a working prototype in under 24 hours Achieved accurate disease identification for common Kenyan crops Built an intuitive interface accessible to non-technical users
What I learned
Effective AI solutions require both technical excellence and a deep understanding of user context. Prompt engineering is as crucial as code engineering.
What's next for Agricheck
Swahili language support Integration with local agricultural extension services Historical tracking of farm health over time SMS-based version for areas with limited internet
Built With
- google-gemini-api
- python
- streamlit
Log in or sign up for Devpost to join the conversation.