DeepLeaf: CNN-Based Precision Agriculture for Automated Potato Leaf Disease Classification

1. training.ipynb (Jupyter Notebook)

• Technologies Used: TensorFlow, Keras
• IDE Used: Jupyter Notebook
• Utilizes TensorFlow and Keras for CNN-based image classification.
• Dataset: PlantVillage.
• Data preprocessing, augmentation, and model training.
• Evaluation and visualization of training/validation metrics.
• Model saving and conversion to TensorFlow Lite format.

2. PlantVillage Dataset

• PlantVillage dataset used for training the CNN model.

Note: Detailed code and data storage specifics are available in the Jupyter Notebook (training.ipynb).

3. Front End: Potato Disease Classification

index.html:

• Main webpage for Potato Disease Classification.
• User-friendly interface for image upload and classification results.
• Responsive design with style customization.

style.css:

• Technologies Used: HTML, CSS
• CSS styling for the front end, enhancing visual appeal and responsiveness.
• Defines layout, colors, and animations for a seamless user experience.

script.js:

• Technologies Used: JavaScript
• JavaScript file handling user interactions.
• Enables image upload, displays a preview, and triggers API calls for disease classification.

4. Additional HTML Pages: Disease Information and Contact

5. FastAPI Backend and Android App Integration:

main.py (FastAPI Backend):

• Technologies Used: FastAPI, TensorFlow
• IDE Used: PyCharm
• Utilizes FastAPI to create an API for potato disease classification.
• Allows CORS for specified frontend URLs.
• Loads a pre-trained TensorFlow model for disease classification.
• Exposes an endpoint "/ping" for a basic health check.
• Exposes an endpoint "/predict" to receive images for classification.
• Returns the predicted class and confidence.

Android App (PotatoPathoGuard in Android Studio):

• Technologies Used: Android Studio, TensorFlow Lite
• Developed using Android Studio.
• Utilizes a TensorFlow Lite (tflite) model produced in the training.ipynb notebook.
• Incorporates functionality to capture or select an image.
• Sends the image to the FastAPI backend "/predict" endpoint for classification.
• Displays the predicted class obtained from the backend.
• Provides a user-friendly interface for interacting with the potato disease classification system.