CNN Visual Learning

This project is an interactive educational application for understanding Convolutional Neural Networks (CNNs). It provides visualizations, comparisons, and hands-on demonstrations of CNNs implemented from scratch and using popular deep learning libraries like TensorFlow and PyTorch.

Features

• Interactive Tutorials: Step-by-step guides to understand CNN concepts.
• CNN Implementations:
  • From scratch using NumPy.
  • Using TensorFlow and PyTorch.
• Visualization Tools:
  • CNN architecture visualization.
  • Convolution and pooling operations.
  • Training process visualization (loss, accuracy, filters, and feature maps).
• Comparison:
  • Performance and code complexity comparison between implementations.
• Custom Filters: Build and apply your own convolutional filters.
• Dataset Support:
  • Preloaded datasets like MNIST.
  • Option to upload custom images.

Installation

1. Clone the repository:

   git clone https://github.com/shivaji-137/CNN-Visual-Learning-.git
   cd CNN-Visual-Learning-

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   streamlit run full_apps.py

Project Structure

.
├── advanced_visualizations.py   # Advanced visualization utilities
├── architecture_builder.py      # Tools for building CNN architectures
├── classification_playground.py # Interactive classification demos
├── cnn_pytorch.py               # PyTorch CNN implementation
├── cnn_scratch.py               # NumPy-based CNN implementation
├── cnn_tensorflow.py            # TensorFlow CNN implementation
├── complete_cnn_copy.py         # Main application logic
├── filters.py                   # Predefined and custom filters
├── full_apps.py                 # Streamlit application entry point
├── guided_tutorials.py          # Step-by-step tutorials
├── training_visualization.py    # Training visualization utilities
├── utils.py                     # Helper functions
├── visualizations.py            # Visualization utilities
├── LICENSE                      # License file
├── README.md                    # Project documentation
├── requirements.txt             # Python dependencies

Usage

1. Launch the application using Streamlit.
2. Use the sidebar to navigate between sections:
   • Introduction: Overview of CNNs.your-username
   • CNN from Scratch: Explore a NumPy-based implementation.
   • Process Sample Images: Apply filters to sample images.
   • Upload Your Own Image: Test CNNs on custom images.
   • Custom Filter Builder: Design and apply custom filters.
   • CNN with Libraries: Explore TensorFlow and PyTorch implementations.
   • Comparison: Compare implementations.
   • Training Visualization: Visualize the training process.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

• Developed by Shivaji Chaulagain.

• Inspired by the need for interactive and visual learning tools for deep learning concepts.

• If you use this code, please give credit to the author.