/ZazuML-easy_AutoML

easy AutoML for Object Detection - Fast-AutoAugment + Hyperband + EfficientDet

Primary LanguagePythonMIT LicenseMIT

Logo

This is an easy open-source AutoML framework for object detection. This project contains an architecture search + hyper-parameter search + augmentations search, as well as trial manager. An ML Pipeline taylor made for simple integration into your project or code.





Algorithms

Augmentations Search Hyper-Parameter Search Architecture Search
  • Fast-Autoaugment
  • Random Search
  • HyperBand
  • EfficientDet
  • EndToEndTransformers

Super EASY to use!

Why ZazuML?

You might be building an ML pipeline to avoid model performance degeneration, or maybe you're just too lazy to download, debug, and tune your own model. Either way, you shouldn't be focusing your efforts on simple things like detection. There's a whole world out there for you to explore, give your hand at trajectory prediction or action recognition and let ZazuML free you up from the boring stuff.

Launch Remotely

ZazuML's REMOTE feature is designed so that you can run search, train and predict anywhere. No GPU's or complex Cloud APIs. It's easy to implement into your python script or run in your terminal.

via Terminal

via iPython

TO DO

  • Increase search space
  • NAS to replace some of the HP search
  • Intelligent Losses to replace some of the HP search

Want to contribute

If you're interested in becoming a collaborator or just have some questions, feel free to join our slack community.

Refrences

Some of the code was inspired by keras-tuner