/MobileNetv2_TF

Tensorflow implementation for Google's MobileNetv2

Primary LanguagePythonApache License 2.0Apache-2.0

MobileNetv2_TF

Introduction

This is a Tensorflow implementation of Google's MobileNetv2. MobileNet is a general architecture and can be used for multiple use cases. Depending on the use case, it can use different input layer size and different head (for example: embeddings, localization and classification). for details, please reference the following paper:

Inverted Residuals and Linear Bottlenecks Mobile Networks for Classification, Detection and Segmentation

MobileNetv2 and inverted residual block architectures

The difference between residual block and inverted residual:

residual block and inverted residual

MobileNetv2 Neural Network:

MobileNetv2

Bottleneck Architectures:

residual block architectures

Requirement

  • python >= 2.7
  • tensorflow >= r1.2

References

Copyright

See LICENSE for details.