About This Project

This system is a capstone prototype for CNN-based tomato disease identification and recommendation. It is designed for smallholder farmers and agricultural students who need a simple web-based tool for early tomato leaf disease screening.

How It Works

1. The user uploads a tomato leaf image.
2. The Flask backend validates and stores the image temporarily.
3. The image is resized to 224 x 224 pixels and normalized.
4. A trained CNN model predicts the disease class.
5. The app displays the result, confidence score, and recommendations.

Limitations

The model is trained using PlantVillage images, which are mostly controlled leaf images. Real farm images may contain different lighting, backgrounds, shadows, blur, or multiple leaves. For field deployment, additional Kenyan farm images should be included during training.