Dataset

The MNIST dataset consists of 60,000 training images and 10,000 test images of handwritten digits (0-9). Each image is 28x28 pixels in grayscale.

Model Architecture (LeNet-5)

• Input Layer: 28×28 grayscale images
• Conv Layer 1: 32 filters (5×5), ReLU activation
• MaxPooling Layer 1: 2×2 pooling
• Conv Layer 2: 64 filters (5×5), ReLU activation
• MaxPooling Layer 2: 2×2 pooling
• Flatten Layer
• Fully Connected Layer 1: 120 neurons, ReLU activation
• Fully Connected Layer 2: 84 neurons, ReLU activation
• Output Layer: 10 neurons (Softmax for classification)

Features

• Implemented LeNet-5 architecture using TensorFlow/Keras
• Trained on the MNIST dataset (28×28 grayscale images)
• Utilizes CNNs for feature extraction and classification
• Evaluates model performance using accuracy metrics
• Developed a user-friendly interface using Streamlit for real-time digit recognition

Getting Started

1. Clone the repository:

   https://github.com/Sam783/LeNet-MNIST.git
   

2. Navigate to the project directory:

   cd LeNet-MNIST
   

3. Install the required dependencies:

   pip install -r requirements.txt
   

4. Run the Streamlit app:

   streamlit run app.py