GSoC 2017 - The End

[Project Website]

Google Summer of Code 2017 has finally come to an end and it couldn’t have been any better. The past three months have been an excellent experience working on Fabrik and it gives me great satisfaction having worked on a project which will potentially benefit a lot of users and researchers in the near future. In the end, I would like to thank my mentors Viraj Prabhu, Deshraj Yadav and Harsh Agarwal for the support and guidance throughout the project.

The project has come a long way from where we started and below is the list of all contributions I have made to it:

Merged Pull Requests

  • PR #56 Continuous integration setup with Travis CI to build latest commits
  • PR #57 Code cleanup for JavaScript files to adhere to recommended guidelines
  • PR #61 Python code formatting to meet PEP 8 standards
  • PR #64 Updated project requirements for developers and added script for Caffe and Tensorflow installation
  • PR #65 & #67 UX changes
    • Changed from a horizontal network graph to a vertical one as scrolling down is more intuitive
    • Added tooltips on hover for nodes(layers) in the network graph
    • Fixed overflow of layer selection dropdown
  • PR #66 & #68 Added guidelines to open issues, submit PRs and contribute to the project
  • PR #70 & #71 & #74 Bug fixes
  • PR #75 Added unit test directory structure and unit tests for all old code
  • PR #76 Changed layer selection dropdown to a hierarchical structure for ease of use
  • PR #77 Support for 35 new Caffe layers to increase Caffe support from 20/56 to 55/56 layers
  • PR #78 Support for Python layer in Caffe which allows users to implement any custom logic with a Python wrapper. This completed the support for Caffe
  • PR #81 Coveralls integration to check and report code coverage
  • PR #82 Unit tests for export logic from Caffe framework
  • PR #83 Initial support for addition of Keras framework to Fabrik
  • PR #85 Export functionality for Keras framework
  • PR #86 & #87 Unit tests for Keras framework and Python layer support
  • PR #88 Bug fixes
  • PR #89 Created and added new logo for Fabrik fabrik_logo
  • PR #91 Support for layers that are exclusive to Keras
  • PR #92 Created a model zoo with predefined networks that can be loaded from a dropdown menu
  • PR #94 Unit tests for Keras export and new layers
  • PR #95 Model sharing feature allowing users to export model to a URL and then load it back at any time
  • PR #96 Complete redesign of the user interface

Evolution of the project UI through the GSoC period Fabrik UI

A run through some features Fabrik UI

Pending Pull Requests

  • PR #93 Support for continuous deployment to automatically push code from master branch to hosting site

Some pull requests were not mentioned as the changes were minor

A more detailed overview of the project is available here