Potato Leaf Disease Classification

Overview

This project aims to develop a machine learning-based classification model for detecting diseases in potato leaves, specifically early blight and late blight. By leveraging deep learning and computer vision, the system assists farmers in early disease diagnosis, enabling timely intervention and minimizing crop loss.

Problem Statement

Potato diseases like early blight and late blight significantly impact crop yield. Traditional disease detection methods are time-consuming and require expert knowledge. This project introduces an automated classification system using deep learning to provide a quick and accurate diagnosis.

Objectives

• Develop a deep learning model to classify potato leaves as healthy or diseased.
• Utilize CNN-based architectures to improve accuracy in image classification.
• Deploy the trained model using Streamlit, making it accessible for farmers.

Methodology

1. Data Collection & Preprocessing: Gather and preprocess images of healthy and diseased potato leaves.
2. Feature Engineering: Apply augmentation techniques to enhance model performance.
3. Model Selection & Training: Train a Convolutional Neural Network (CNN) for classification.
4. Evaluation: Assess model performance using accuracy, precision, recall, and F1-score.
5. Deployment: Deploy the trained model using Streamlit, allowing real-time image classification.

Software & Hardware Requirements

Hardware:

• Processor: Intel i5/Ryzen 5 or higher
• GPU: NVIDIA GTX 1060 or better (recommended for training)
• RAM: 16GB (32GB recommended)
• Storage: At least 100GB free SSD space

Software:

• Operating System: Windows 10/11, macOS, or Linux
• Programming Language: Python 3.10+
• Libraries:
  • TensorFlow / PyTorch
  • Streamlit
  • Pillow
  • NumPy
  • Matplotlib
  • Seaborn
• Development Tools:
  • Jupyter Notebook / VS Code / PyCharm
  • Git for version control

Results & Performance

• The trained model achieved high accuracy in distinguishing healthy vs. diseased leaves.
• The Streamlit interface allows users to upload leaf images and get real-time predictions.

Future Work

• Expand dataset for better model generalization.
• Improve model accuracy with advanced architectures (e.g., EfficientNet, ViTs).
• Deploy as a mobile app using TensorFlow Lite.
• Integrate weather and environmental data to improve prediction reliability.

Contributors

• S Yogesh Reddy – Project Development
• Jay Rathod and Dr. V Keerthika – Guidance & Mentorship

Acknowledgment

We extend our gratitude to our supervisor, faculty members, and fellow researchers for their support and insights throughout this project.