search

havanagrawal / clomask

Capstone Project for Clobotics: Using Mask R-CNN for Rigid/Non-Rigid Retail Consumable Product Detection
MIT License
7 stars 4 forks source link

Synthetic Data Generation Module #22

Closed vivanvish closed 5 years ago

vivanvish commented 5 years ago

This PR contains the changes for the Synthetic Data Generation Module. Refer to the Demo.ipynb notebook on how to use the module and as well the prerequisites.

The module needs the pillow python package.

havanagrawal commented 5 years ago

Can we not check in generated artifacts please (generated image and masks).

havanagrawal commented 5 years ago

Going forward, I would also recommend rebasing changes from master into a feature branch which keeps the commit history clean, instead of merging which creates an unnecessary "merge commit".

vivanvish commented 5 years ago

I decided to check in the generated artifacts to keep them as reference sample.

havanagrawal commented 5 years ago

Can we instead display the masks in the demo notebook itself, and clean up the artifacts after that?

If not, it would probably be better to have a smaller example with less than ten masks. Committing 100 synthetic images into the repo feels like overkill to me tbh

havanagrawal commented 5 years ago

Comments for the notebook:

  1. Cell 28 has duplicated pydocs. If we update the class or method, we'll have to update the cell as well. Instead, consider using the help method, or the __doc__ property, both of which will show the documentation: 
    help(synthesizer.generate_synthetic_dataset)
synthesizer.generate_synthetic_dataset.__doc__