Banana Leaf Disease Detection

Datasets Used

The model was trained on a combined dataset from two sources:

1. Banana Leaf Disease Dataset - Kaggle
2. Mendeley Banana Leaf Dataset

the combined dataset includes the following 7 classes:

• Sigatoka
• Moko
• Insect Pest
• pestalotiopsis
• Bract Mosaic Virus
• cordana
• Healthy

Models Used

The following CNN architectures were fine-tuned for classification:

• VGG16
• ResNet152V2
• InceptionV3
• MobileNetV2
• LeNet (custom implementation)

The best performing model - VGG16 is saved as model.h5 for inference in the web application.

Workflow of Proposed Model

Setup

Follow these steps to set up the project on your local machine.

1. Clone the Repository

git clone https://github.com/Manoj632004/Banana_plant_decease_detection
cd banana-leaf-disease-detection

2. Install Dependencies

pip install -r requirements.txt

3. Train Model

• Run the run.ipynb notebook to train your desired CNN model.
• Export the model as model.h5

4. Set Up Static Folder for Uploads

mkdir static

5. Run the Flask App

python app.py