l4es / tf-serving-client-cpp

An example of TensorFlow Serving client to send images for prediction (using C++ and CMake)

This project is a kind of boilerplate to be used to get acquainted with a process of TensorFlow Serving client development using C++ as a programming language and CMake as a build system.

PREREQUISITES

To use a TensorFlow Serving client provided in this project you should install next dependencies:

• Protocol Buffers (https://github.com/google/protobuf/tree/master/src):

  • sudo apt-get install autoconf automake libtool curl make g++ unzip
  • git clone https://github.com/google/protobuf
  • cd protobuf
  • ./autogen.sh
  • ./configure && make && make check && sudo make install && sudo ldconfig

• gRPC (https://github.com/grpc/grpc/tree/master/src/cpp):

  • sudo apt-get install build-essential autoconf libtool
  • git clone -b $(curl -L https://grpc.io/release) https://github.com/grpc/grpc
  • cd grpc
  • git submodule update --init
  • make
  • sudo make install

• OpenCV (http://opencv.org/), there are a lot of ways of the installation are available.

COMPILATION AND USAGE

Before the compilation make sure that you have provided some host with a running TensorFlow Serving model in the main.cpp file.

The compilation is quite similar to ones which use CMake system. But make sure to provide a path to the installed Protobuf Compiler using Protobuf_PROTOC_EXECUTABLE for it. For example you can use a bunch of commands (from the root of the repo):

mkdir build && cd build && cmake -D Protobuf_PROTOC_EXECUTABLE=/usr/local/bin/protoc .. && make

A built executable accepts one argument with a path to an image to be sent to the server:

./tf-serving-client-cpp test.jpg