/keras-shufflenet

ShuffleNet Implementation using Keras Functional Framework 2.0

Primary LanguagePythonMIT LicenseMIT

Keras ShuffleNet Build Status

ShuffleNet Implementation using Keras Functional Framework 2.0

There is a weight file for groups=3 but I was not able to reproduce the accuracy from the paper. Current accuracy for ImageNet is 0.5228.

Library Versions

  • Keras v2.0.8+
  • Tensorflow 1.0+

References

  1. Keras Framework

  2. ShuffleNet Paper

Licence

MIT License

Note: If you find this project useful, please include reference link in your work.