Multi-Dataset Image Classification (LeNet Streamlit)

This project provides an implementation of image classification for multiple datasets using the LeNet architecture. It supports MNIST and Cassava Leaf Disease datasets and includes a Streamlit-based user interface for convenient interaction.

Features

• LeNet Model: Implements two variations of the LeNet architecture for grayscale (MNIST) and RGB (Cassava) images.
• Dataset Support:
  • MNIST: Classifies handwritten digits.
  • Cassava Leaf Disease: Identifies various diseases in cassava leaves and healthy leaves.
• Streamlit UI: A simple web application to classify uploaded or example images.

Prerequisites

Ensure the following are installed:

• Python 3.7+
• Required Python libraries. (in requirements.txt)
• Pretrained model weights:
  • MNIST: lenet_model_mnist.pt
  • Cassava: lenet_model_cassava.pt

Installation

1. Clone the repository:

   git clone https://github.com/OxyzGiaHuy/LeNet_Streamlit.git
   cd LeNet_Streamlit

2. Install dependencies:

   pip install -r requirements.txt

Usage

Run the Application

Start the Streamlit application with the following command:

streamlit run app.py

User Interface

1. Dataset Selection:
   • Choose between MNIST and Cassava Leaf Disease in the sidebar.
2. Input Options:
   • Upload an image file (JPG/PNG).
   • Run an example image provided in the repository.
3. Output:
   • Displays the image.
   • Shows the predicted label and confidence percentage.

Model Details

LeNet for MNIST

• Input: Grayscale images (28x28)
• Layers:
  • Conv2D (1 -> 6)
  • Average Pooling
  • Conv2D (6 -> 16)
  • Average Pooling
  • Fully Connected Layers

LeNet for Cassava Leaf

• Input: RGB images (150x150)
• Layers:
  • Conv2D (3 -> 6)
  • Average Pooling
  • Conv2D (6 -> 16)
  • Average Pooling
  • Fully Connected Layers

Directory Structure

LeNet_Streamlit/
|-- requirements.txt
|-- app.py                     # Main application script
|-- model/
|   |-- lenet_model_mnist.pt   # Pretrained MNIST model
|   |-- lenet_model_cassava.pt # Pretrained Cassava model
|-- example/
    |-- demo_8.png             # Example MNIST image
    |-- demo_cbsd.jpg          # Example Cassava image

Future Enhancements

• Support for additional datasets.
• Integration with cloud services for model hosting.