Skip to content

Latest commit

 

History

History
137 lines (98 loc) · 4.62 KB

README.md

File metadata and controls

137 lines (98 loc) · 4.62 KB
Raw

ez-nn (Easy Neural Networks)

An easy graphical user interface for creating image based convolutional neural networks using the TensorFlow library.

Current Release

For Feature Suggestions and Bug Reports

Please use the issues tracker on this GitHub repository!

Current Features

  • Create image neural networks and test them using this easy to use GUI.
  • Actual applicability of this application is likely limited, this is just meant for a fun demonstration of TensorFlow.

Upcoming Features (higher importance closer to the top)

  • Lots of commenting and documentation
  • "Behind the scenes" documentation
  • Code shortening and increased efficiency
  • Cleanup confusing mix of classes and ids
  • TensorBoard Guide
  • Support for .png images
  • Possible stopping of TensorBoard inside the application

Known Issues

  • TensorBoard does not interact properly on windows
  • Spaces in folder names or subfolder names cause problems with training and testing pictures

Prerequisites

  • Internet - This application requires an active internet connection

  • Linux: Fully functional, Tested on Ubuntu 16.04.3 LTS

or

or

  • MacOS: Likely functional but not tested

Installations

Installing and Deployment

  1. Clone the repository.
  2. Install Python ^3.6 or use Anaconda if on windows, make sure you install to PATH environment variable.
  3. Install TensorFlow. TensorFlow must be installed in a separate environment called tensorflow. Also make sure to follow all the directions very carefully and install in the way Google recommends.
  4. Install node.js/npm
  5. Install yarn
  6. In the directory of the repository execute the command: yarn install.
  7. Then yarn start.
  8. Done!
  9. If errors occur try installing under root and switching back to user before start. It is currently unknown why this is sometimes necessary.

Most editing you will want to do if any will take place in tf.js.

Happy forking!

Users' Guide

View the wiki page here.

Troubleshooting

View the wiki page here.

Built With

Pip Packages

  • TensorFlow 1.5.1
    • It should be noted that lots of the backend this project uses is provided by the wonderful TensorFlow library which it itself is licensed under the Apache License, Version 2.0 by Google Inc.

Node Modules

Contributing

Submit an issue tracker or a pull request and someone will get back to you.

Authors

Any future collaborators will be added here.

License

This project is licensed under the GNU GENERAL PUBLIC LICENSE (Version 3, 29 June 2007) - see the LICENSE file for details.

DISCLAIMER: This license only applies to material not linked above! To find the license of each tool/language/api/library ez-nn is created with, please view the links provided for each one as listed above.

Acknowledgments

  • Thanks to Google for the awesome TensorFlow API library :)