• 

  App landing screen before a farmer can capture an image of banana plantation leaf

Rwanda's AI-Powered Solution for Banana Disease Detection

Inspiration

Banana farming is vital in Rwanda but faces massive losses due to diseases like Panama Disease and Black Sigatoka. Existing manual methods are slow and ineffective. We wanted to create an easy, tech-based solution to help farmers detect and manage these diseases quickly.

What it does

The app uses AI and machine learning to:

• Detect banana diseases through photos.
  
• Classify diseases accurately.
  
• Recommend treatments tailored to each disease.
  
• Provide a multilingual interface to ensure accessibility for all farmers.

How we built it

1. Data Collection: Took photos of banana plants from farms and expanded the dataset with augmentation techniques.
2. Model Training: Used CNNs with PyTorch and TensorFlow Lite to train and optimize models for mobile use.
3. App Development: Built a lightweight Android app to process images and provide results.
4. Testing: Ensured the system works with over 90% accuracy.

Challenges we ran into

• Limited local datasets.
• Optimizing AI models for mobile performance.
• Making the app user-friendly for farmers.
• Differentiating diseases with similar symptoms.

Accomplishments that we're proud of

• Created a high-accuracy AI model for detecting banana diseases.
• Built a simple, multilingual app accessible to rural farmers.
• Developed a solution tailored to Rwanda's farming challenges.

What we learned

• AI can transform farming by making disease detection faster and more accurate.
• Localized solutions are key to solving real-world problems.
• Involving farmers in design makes the solution practical and user-friendly.

What's next

• Collect more data to improve the model.
• Make the app faster and easier to use.
• Expand the technology to detect other crop diseases.
• Partner with organizations to bring the solution to more farmers.

Built With

• python
• react-native