Skip to content
This repository has been archived by the owner on Feb 17, 2021. It is now read-only.

Keras implementation of Squeeze Net architecture described in arXiv : 1602.07360v3 (https://arxiv.org/pdf/1602.07360.pdf)

License

Notifications You must be signed in to change notification settings

AlexandruBurlacu/keras_squeezenet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

58 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Keras Squeezenet

Keras implementation of Squeeze Net architecture described in arXiv : 1602.07360v3 (https://arxiv.org/pdf/1602.07360.pdf)

It uses the so-called Eve optimizer (https://arxiv.org/pdf/1611.01505v2.pdf), implementation here: https://github.com/jayanthkoushik/sgd-feedback

TODOs:

  • add data augmentation (Done)
  • add bypasses (Added simple bypasses)
  • experiment with Batch Normalization (Parametrized inside SqueezeNetBuilder and FireModule classes)
  • try more levels of Dense layers (Added the possibility to inject a small subnet in the model)
  • Update Eve Optimizer implementation to Keras 2

Model's Graph (using bypasses)

About

Keras implementation of Squeeze Net architecture described in arXiv : 1602.07360v3 (https://arxiv.org/pdf/1602.07360.pdf)

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published