Final YearAdvancedAI/ML 3-4 months
Crop Disease Detection (Mobile CNN)
A mobile app for farmers that identifies plant diseases from leaf photos using an on-device CNN and suggests treatment, working offline.
PythonTensorFlow / KerasTensorFlow LiteFlutter
Find GitHub repos & source code →How to build it — step by step
- 1Dataset + training: Train a CNN (transfer learning) on a leaf-disease dataset with augmentation.
- 2On-device model: Quantise and convert the model to TensorFlow Lite for offline mobile inference.
- 3App: Capture/upload a leaf image, run inference, and show disease + confidence.
- 4Guidance: Map each disease to concise, sourced treatment and prevention advice.
Key features to implement
- ✓Offline on-device inference
- ✓Leaf-photo disease detection
- ✓Confidence scores
- ✓Treatment guidance
- ✓Multi-crop support
💡 Unique twist to stand out
Add a severity estimate from the affected leaf area and a community feature to report local outbreaks on a map.
🎓 What you'll learn
Transfer learning, model quantisation, on-device ML, and mobile deployment.