Skip to content
This repository has been archived by the owner on Oct 11, 2019. It is now read-only.
/ from-keras-to-c Public archive

A practical example of Tensorflow C API based deployment starting from a model trained with Tensorflow + Keras

License

Notifications You must be signed in to change notification settings

aljabr0/from-keras-to-c

Repository files navigation

From Keras to C

This small demo project is about deploying deep learning models on embedded platforms. The techniques exposed here have been particularly useful to me in the deployment of deep learning models in industrial applications.

We start with a simple example model, trained with Tensorflow 1.x + Keras. In the end, we'll freeze the model and export a GraphDef that can be loaded and executed through the Tensorflow C API (without Python). The training and export are coded into the file train_and_export.py whereas the inference is coded into the file model_run.cpp. The Tensorflow binaries must be loaded into the folder lib, they can be downloaded from the following link.

Note Once the API for Tensorflow 2.0 will be stable the project will be updated to the newest version of the framework.

About

A practical example of Tensorflow C API based deployment starting from a model trained with Tensorflow + Keras

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published