Denisganga / RustAndYellowMosaicDiseaseDetection

a model to identify the yellow mosaic and rust disease on greengrams

Plant Doctor - Yellow Mosaic and Rust Disease Detection on Green Grams

Overview

This project demonstrates the process of building a simple image classification model using PyTorch. The model is designed to classify plant diseases, with a focus on detecting yellow mosaic and rust diseases on green grams.

Requirements

Before running the project, ensure you have the following dependencies installed:

• PyTorch
• torchvision
• Google Colab (if running on Colab)
• Matplotlib
• Pillow

Usage

1. Mount Google Drive:

   • Mount your Google Drive to access the dataset stored there. The dataset is not a complicated and it got few datasets. The dataset is already included on this repossitory with a folder plant_doctor

2. Set the path to your dataset within Google Drive:

   • Update the data_dir variable with the path to the plant_doctor dataset on your Google Drive.

3. Define Image Transformations for Preprocessing:

   • The transform variable contains the required transformations for image preprocessing.

4. Create a Dataset and Split into Training and Testing Sets:

   • The dataset is created using the ImageFolder class, assuming each subfolder in data_dir represents a different class.
   • The dataset is split into training and testing sets.

5. Create a DataLoader for Efficient Data Loading:

   • Utilize DataLoader to efficiently load and batch the data.

6. Define and Train the Baseline CNN Model:

   • A baseline CNN model (BaselineModel) is defined and trained using a specified criterion and optimizer.

7. Evaluate the Model on the Test Dataset:

   • The model's performance is evaluated on the test dataset, and the test accuracy is printed.

8. Save the Trained Model:

   • The trained model is saved as 'baseline_model.pth' for reuse in predictions or further training.

9. Load the Model for Further Use:

   • The model architecture is loaded, and its weights are loaded from the saved file.

10. Make Predictions on a New Image:

    • A new image is preprocessed, passed through the loaded model, and its predicted class label and confidence are printed.

Note

• Ensure that your dataset is structured in subfolders, each representing a different class.
• The provided model architecture (BaselineModel) is a simple example and may need adjustments based on the complexity of the dataset.

Feel free to customize the project based on your specific dataset and requirements.