Firimoni

What it does

My application leverages Gemini's natural language capabilities to empower marginalized/non-marginalized farmers. Using sensors that collect real-time data and feed it to Firebase, the system monitors specific plots or hydroponic tanks. This data, accessible to the Gemini API, is integrated with a WhatsApp interface to reach farmers who may not have access to advanced AI tools but regularly use WhatsApp.

The system includes three CNN-trained TensorFlow models capable of classifying four tomato diseases, three potato diseases, and six maize diseases. Farmers interact with these models through WhatsApp. When a disease is predicted, the model provides a class and confidence level, which Gemini translates into understandable language. Gemini engages with farmers in a friendly manner, offering insights to improve crop health, reduce pesticide use, and explore alternative farming methods.

I've also implemented Retrieval-Augmented Generation (RAG) capabilities, embedding PDF data on farming practices and crop management. When the agent is not confident in an answer, it uses Exa to search the internet for accurate information. By maintaining context about each farmer's plot through Firebase IoT data, Gemini offers personalized recommendations, helping farmers grow more nutritious food and supporting sustainable practices.