Plant Disease Prediction
🌿 Deep learning · Runs in your browser

Spot leaf disease in seconds.

Upload a photo of a leaf and a convolutional neural network tells you whether it looks healthy, or shows signs of powdery mildew or rust.

  • 3classes
  • 1,322training images
  • 0uploads to a server
Close-up of a green leaf

Upload a leaf photo

Or try a sample:
Loading model…

This is a student machine-learning project for learning purposes. Predictions can be wrong — consult an agronomist before treating crops.

What the model looks for

Healthy leaf

Healthy

Even green colour, no white powder, spots or pustules on the leaf surface.

Leaf with powdery mildew

Powdery mildew

A fungal disease that leaves white or grey powdery patches, usually starting on the upper side of leaves.

Leaf with rust

Rust

Fungal infection showing orange, yellow or brown pustules that release rust-coloured spores.

How it works

  1. Resize. Your photo is scaled to 225 × 225 pixels and normalised to 0–1.
  2. Extract features. Two convolution + max-pooling blocks learn edges, textures and spots.
  3. Classify. Dense layers turn those features into probabilities for the three classes.
  4. Private by design. The model runs with TensorFlow.js on your device — the image is never uploaded.
Input225×225×3
Conv2D 32 · ReLU223×223×32
MaxPool 2×2111×111×32
Conv2D 64 · ReLU109×109×64
MaxPool 2×254×54×64
Flatten186,624
Dense 64 · ReLU64
Dense 3 · Softmax3