Leaf disease detection from phone photos

The brief

Farmers lose crops because diseases are spotted late. Train an image classifier on leaf images, then test it on real field photos, which are much harder than lab images, and show where it fails. Good looks like: accuracy per disease, an honest test on field photos, and advice that sends unsure cases to an expert.

What you will build

1. Train on lab images of healthy and diseased leaves
2. Test on field photos and measure the drop
3. Fine-tune with field images and augmentation
4. Evaluate per disease
5. Build a phone-friendly demo with confidence scores

Free, open data to build it with

• PlantVillage dataset ↗ Kaggle
• PlantDoc field images ↗ GitHub

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.