Farm-AI

The application UI
downloaded an image with leaf spot disease for uploading to our app.
Uploaded the image and the AI analyzes it.
I asked a question on how to prevent and treat the leaf spot disease, and the AI generated a response.
The AI tool returned an accurate response

Inspiration

The inspiration for Farm-AI came from the struggles of small-scale farmers who often face crop diseases without access to timely expertise. Many of these farmers are unable to identify diseases early on, leading to loss of crops and reduced yields. By providing an accessible AI tool, Farm-AI aims to empower local farmers with the ability to diagnose crop diseases through uploading images of infected crops, and the farmers can gain valuable insights on treatment and prevention, and this will ultimately improve crop health and productivity.

What it does

Farm-AI enables farmers to upload images of infected crops for analysis. The AI processes the image with advanced AI capabilities, identifies potential diseases, and provides a diagnosis. Farmers can also ask additional questions using the AI ChatBox, about disease prevention and treatment methods. This AI tool has impressive accuracy, and will help farmers make informed decisions about crop management, reducing the risk of crop loss and improving crop yield.

How we built it

Farm-AI was built using the an API from Gemini to power image analysis and question answering. We used a web-based interface for easy access, allowing farmers to upload images and input questions. We used Google AI Studio was used for training the model on crop disease detection, while Vertex AI handles the deployment, ensuring a smooth and scalable experience. We used the following - Technologies:

JavaScript, HTML and CSS: which form the front end, creating a user-friendly interface for local farmers.
Backend script: Integrated JavaScript and harness the power of Gemini API
nix: a cross-platform package manager for Unix-like systems, and a tool to instantiate and manage those systems
Checkout the project-repo for more details.

Challenges we ran into

The challenges we encountered included:

Training the AI to recognize a wide variety of crop diseases, to ensure accuracy even with limited image quality.
Optimizing the platform to function well even in areas with low internet connectivity.

Additionally, making the interface easy for farmers with varying levels of tech familiarity and knowledge was a key focus area, which was also quite challenging but doable.

Accomplishments that we're proud of

We’re proud to have developed an accessible tool that empowers local farmers with AI-driven insights. Our accomplishments includes:

Successfully integrating image analysis with question-answering capabilities.
Achieving high accuracy in disease detection.
Creating a simple UI with an intuitive user experience that can be easily used by non-technical users.

What we learned

Working on Farm-AI helped us deepen our understanding on building a powerful AI Image Analyzer that analyzes images with advanced AI capabilities, and a very impressive accuracy. We also learned how to integrate Gemini API which is the engine that powers our entire application. This was a new concept for us, but we took on the challenge and learned a lot in the process, gaining valuable experience in deploying AI models in real-world applications. Additionally, we got to research and know the challenges local farmers face in the field.

What's next for Farm-AI

There is still plenty of room for improvement. We plan to:

Expand Farm-AI’s disease recognition database to cover more crops and disease types.
Add recognition of animal/livestock diseases, ensuring that it serves a broader range of farmers.
Improve accuracy further and explore offline functionality for farmers with limited internet access.