← all projects
case study

AI Plant Disease Detector

A fine-tuned EfficientNet CNN that diagnoses plant diseases from a photo — 98.6% accuracy across 38 classes — wrapped in a Django web app.

  • TensorFlow
  • Python
  • Django
  • HTML

Can a photo tell you why your tomato plant is dying? Usually, yes. I fine-tuned an EfficientNet-based CNN with TensorFlow to classify plant species and their diseases, then wrapped it in a Django web app so the answer is one upload away.

The model

Trained and validated on a custom dataset of 87K RGB images across 38 classes of species and diseases, reaching 98.6% classification accuracy while generalizing well across categories. It's deliberately lightweight — the goal was something that runs cheaply in real applications, not a leaderboard entry.

The app

A full-stack inference pipeline built with Django and HTML: upload a photo, get the classification in real time, and keep a history of past results — responsive and friendly enough to hand to someone who just wants their plant fixed.

Browse the other projects