TensorFlow Lite for Go

Overview

TensorFlow Lite known as TFLite, is an open source deep learning framework for on-device inference. Current API supports officially python, java and C++. Our project creates and proposes a binding for Golang using internally TensorFlow Lite C API, providing an smooth Go experience.

tflitego provide a simple and clear solution to use TensorFlow lite in Go. Our objective is provide a cohesive API, simplicity related to TensorFlow Lite C API connection and maintainability.

repository: https://github.com/nbortolotti/tflitego

Dependencies

TensorFlow Lite C library

I prepared some packages to provide TF Lite C Library installs simplification.

raspberrypi_linux/ARMv7, 2.4.0

Note: all the elements added into Home https://github.com/nbortolotti/tflitego

Go Language

Go version (tested): 1.15

Binding Methodology

To create the binding from C to Go, we are using cgo

Note: more information related to design and binding created into Design section

How to use?

Here I prepared a section with examples and the generic model to use the library.