Maize Leaf Disease Detection using Deep Learning

This project implements an end-to-end deep learning approach to automatically detect and classify maize (corn) leaf diseases from images using a Convolutional Neural Network (CNN).

The model classifies maize leaf images into four categories: Blight, Common Rust, Gray Leaf Spot, and Healthy.

Problem Statement

Maize leaf diseases significantly reduce crop yield and food security. Traditional disease identification relies on manual inspection by experts, which is time-consuming and not scalable.
This project aims to automate maize leaf disease detection using deep learning techniques.

Dataset

• Source: Kaggle – Corn or Maize Leaf Disease Dataset
• Total Images: 4,188
• Classes:
  • Blight
  • Common Rust
  • Gray Leaf Spot
  • Healthy

The dataset was divided into training, validation, and test sets.

Methodology

1. Data cleaning and validation
2. Train–Validation–Test split
3. Image preprocessing (resize and normalization)
4. Model training using ResNet-18
5. Model evaluation using accuracy and confusion matrix
6. Prediction on unseen images

Model Details

• Architecture: ResNet-18
• Framework: PyTorch
• Input Size: 224 × 224
• Loss Function: Cross-Entropy Loss
• Optimizer: Adam

Results

• Training Accuracy: ~99%
• Test Accuracy: ~97–99%
• High precision and recall across all disease classes

The model successfully predicts disease categories for unseen maize leaf images.

Project Structure