Skip to content

README.md

Latest commit

 

History

History
94 lines (69 loc) · 3.45 KB

README.md

File metadata and controls

94 lines (69 loc) · 3.45 KB

Fruits and Vegetables Disease Detection 🍎🥦 (As an example)

A deep learning project to classify diseases in fruits and vegetables using CNNs and traditional ML models (SVM, KNN, Random Forest).

Note: This project is designed to be adaptable for any CNN-based classification dataset, making it a versatile tool for various image classification tasks.

Demo Python

Table of Contents

Project Overview

This project aims to:

  1. Detect Diseases: Classify fruits/vegetables as healthy or diseased (e.g., Apple_Healthy vs. Apple_Rotten).
  2. Compare Models: Evaluate CNN performance against traditional ML models (SVM, KNN, Random Forest).

Key Features:

  • Data preprocessing and augmentation.
  • CNN model training with TensorFlow/Keras.
  • Traditional ML pipelines with scikit-learn.
  • Model accuracy comparison and visualization.
  • Confusion matrix and classification report generation.

Dataset

The dataset is downloaded from Kaggle:
Fruit and Vegetable Disease Dataset

Structure:

  • Images of fruits and vegetables categorized as Healthy or Diseased.
  • Split into training, validation, and test sets during preprocessing.

Installation

Follow these steps to set up the project:

  1. Clone the Repository: 
    git clone https://github.com/SecurDrgorP/Model_Crafter_Project.git
cd Fruits-and-Vegetables-Disease-Detection

Usage

To run the project, follow these steps:

  1. Prepare the Dataset:

    • Ensure the dataset is downloaded and placed in the data/raw directory.
    • Run the main script to clean and preprocess the dataset: 
      python main.py

  2. Train the CNN Model:

    • During training, you will be prompted to choose whether to use the custom checkpoint logic: 
      Do you want to use the custom checkpoint logic? (y/n):
    • Type y to enable saving the model based on the lowest difference between training and validation accuracy and the lowest validation loss.

  3. Evaluate Models:

    • The pipeline will automatically evaluate both CNN and traditional ML models and save the results in the results/ directory.

  4. View Results:

    • Check the classification reports, confusion matrix, and model comparison CSV in the results/ directory.

Features

  • Custom Checkpoint Logic: Save the CNN model based on the lowest difference between training and validation accuracy and the lowest validation loss.
  • Traditional ML Models: Compare CNN performance with SVM, KNN, and Random Forest.
  • Visualization: Generate confusion matrices and classification reports for better insights.
  • Data Augmentation: Automatically applies augmentation to improve model generalization.

Results

  • CNN Model: Achieved high accuracy in detecting diseases in fruits and vegetables.
  • Traditional ML Models: Performance varies depending on the dataset and preprocessing.
  • Comparison: Results are saved in results/f1_comparison.png for easy analysis.