Cassava Leaf Disease Identification

A Deep Learning Approach

Here's a simple README for your GitHub project:

---

This project identifies diseases in cassava leaves using a deep learning model based on the pre-trained ResNet50 architecture. The project includes two user roles: Farmer and Agricultural Officer.

Setup Instructions

1. Install Required Libraries:

   • TensorFlow (for deep learning)
   • Pillow (for image loading)
   • Flask (for the web framework)

   Run the following command to install them:

   pip install tensorflow pillow flask

2. Database:

   • The project uses SQLite3 as the database. No need for external setup.

3. Run the Project:

   • First, set up the database by running:

     python create.py

   • Then, start the application:

     python app.py

Project Overview

• Farmer Module: Allows farmers to log in and upload images of cassava leaves for disease identification. The system predicts the disease and provides prevention methods.

• Agricultural Officer Module: Officers can log in to manage disease data, including adding new diseases and updating prevention methods.

Abstract

Cassava plants are vital for food security in Africa, but viral diseases threaten crop yields. This project trains a CNN model on ~21k images to classify 4 disease types and healthy plants. Using transfer learning with ResNet50, we achieved a validation accuracy of ~86%. Future work could involve Test Time Augmentation and experimenting with larger image resolutions and other architectures.