Midas Green, Democratizing Agriculture, Nurturing Growth

Inspiration

We got inspiration for this project from our experience volunteering with the Santa Cruz Fruit Growers. Growing densely planted fruit crops proliferated disease, and we spent days diagnosing powdery mildew, mite damage, and anthracnose on the plants we grew.

What it does

Midas Green helps democratize access to agricultural technology by identifying plant diseases

How we built it

Midas Green utilizes a Convolutional Neural Network (CNN) and fine-tuned Gemma model optimized with Intel INT8 quantization and RAG to ensure accurate responses. With our CNN, we currently can classify 32 plant diseases from images captured in any garden. The backend, hosted on Intel Cloud VM, efficiently processes the data, while the development environment on Intel Cloud ensures seamless integration and scalability. The model outputs disease classifications, which are then analyzed through a Retrieval Augmented Generator (RAG) to suggest effective treatment options for farmers.

Challenges we ran into

Creating our own CNN and fine tuning the Gemma model was difficult

Accomplishments that we're proud of

We're really proud of how well Midas Green uses modern AI technology, and how much we've learned about LLM's and deep learning along the way

What we learned

Each developer embarked on a journey of learning and growth, as we all worked with technology that we hadn't previously React, Node.js, and Large Language Models. Our exploration led us to discover the remarkable capabilities of React and Node.js in building robust, scalable web applications. With React's component-based architecture and Node.js's event-driven, non-blocking I/O model, we found ourselves empowered to create dynamic and efficient user experiences.

What's next for Midas Green

Moving forward, we envision expanding the capabilities of Midas Green to cover a broader range of crops and diseases. Furthermore, we aim to partner with agricultural companies by verifying quality products to recommend to gardeners.

Built With

Share this project:

Updates