README.md

metadata

license: gpl-3.0

Potato and Tomato Disease Classification Web Application

Overview

This project is a web application developed using Flask that allows users to upload images of potato or tomato leaves and receive predictions regarding potential diseases. The application utilizes two deep learning models: one trained to classify potato leaf diseases and another for tomato leaf diseases. Both models were trained using convolutional neural networks (CNNs) and implemented using PyTorch.

Key Features

• Image Upload: Users can upload images of potato or tomato leaves.
• Disease Prediction: The application predicts whether the leaf is healthy or affected by specific diseases.
• Dynamic Background: The background image of the web page dynamically changes based on whether the user selects potato or tomato.
• Probability Display: The probability of the predicted class is displayed as a percentage.

Technologies Used

• Python: Core programming language used for model development and Flask backend.
• Flask: Web framework for developing the web application.
• PyTorch: Deep learning framework used to develop and train the models.
• HTML/CSS: For creating the frontend of the web application.
• PIL (Pillow): For image processing.
• OpenCV: For image display and preprocessing.
• Torchvision: For image transformation utilities.

Models

• Potato Disease Classification Model

  • Classes: Potato Early Blight Potato Late Blight Potato Healthy
  • Techniques Used: Convolutional layers for feature extraction. Batch normalization and max pooling for enhanced training stability and performance. Dropout layers to prevent overfitting.

• Tomato Disease Classification Model

  • Classes: Tomato Early Blight Tomato Late Blight Tomato Healthy
  • Techniques Used: Similar architecture to the potato model with appropriate adjustments for tomato disease classification. Batch normalization, max pooling, and dropout layers are also used here.

Web Application

• Backend The backend of the application is powered by Flask. It loads the trained models, handles image uploads, processes images, and returns predictions to the frontend.

  • Model Loading: Both models (potato and tomato) are loaded at the start of the application to minimize latency during prediction.
  • Prediction Logic: Depending on the selected plant type (potato or tomato), the corresponding model is used to predict the disease class.
  • Dynamic Background: The background image on the frontend changes based on the selected plant type, enhancing user experience.

• Frontend The frontend is developed using HTML and CSS, with Bootstrap for responsive design.

  • File Upload Interface: Users can upload an image of a leaf.
  • Prediction Display: After processing, the application displays the predicted disease class and the associated probability.
  • Dynamic Background: The background image changes depending on whether the user is predicting for potato or tomato.

Usage

1. Install the required dependencies using pip install -r requirements.txt.
2. Download the pre-trained model weights and place them in the models/ directory.
3. Run the Flask web application using python app.py.
4. Access the application in your web browser at http://localhost:5000.

Outcome

• Performance
  • Potato Model: Achieved an accuracy of 98% on the validation set, with strong performance in classifying Early Blight, Late Blight, and Healthy leaves.
  • Tomato Model: Achieved an accuracy of 97% on the validation set, effectively distinguishing between Early Blight, Late Blight, and Healthy leaves.
• Benefits
  • Disease Detection: Helps farmers and agriculturists detect diseases in potato and tomato plants early, potentially preventing crop losses.
  • User-Friendly Interface: The web application provides a simple interface for non-technical users to diagnose plant diseases.