Android App With Tflite C++ Api

Android App using Tflite C++ API

In this blog, I’ll show you how to build an Android app that uses Tflite C++ API for loading and running tflite models. Please note that this tutorial assumes you are using Ubuntu 16.04 system.

Building Tflite shared library

We’ll build Tflite shared library from tensorflow sources. Following steps illustrate how to build Tflite shared library from sources.

Get tensorflow code

mkdir -p $HOME/tf
cd $HOME/tf
git clone https://github.com/tensorflow/tensorflow.git

Install Bazel to build tensorflow

Follow instructions from https://docs.bazel.build/versions/master/install-ubuntu.html.

Install Tensorflow from Sources

Follow instructions from https://www.tensorflow.org/install/install_sources

Build Tflite

• Edit $HOME/tf/tensorflow/tensorflow/contrib/lite/BUILD. Add the following code to the end of the file:

# Build Tensorflowlite Library
cc_binary(
    name = "libtflite.so",
    deps = [
        ":framework",
        "//tensorflow/contrib/lite/kernels:builtin_ops",
    ],
    linkshared=1
)

• From shell, run the following commands:

cd $HOME/tf/tensorflow/
bazel build -c opt //tensorflow/contrib/lite:libtflite.so --crosstool_top=//external:android/crosstool --host_crosstool_top=@bazel_tools//tools/cpp:toolchain --config=android_arm64 --cpu=arm64-v8a --fat_apk_cpu=arm64-v8a --cxxopt="-std=c++14"

• Copy tflite.so file from /home/sanjay/tf/tensorflow/bazel-bin\* folder to a local folder.

Integrating libtflite.so with Android app

I’m using Android Studio version 3.14 and Android NDK Version 17.1

Make changes to CmakeLists.txt

set(HOME_DIR                $ENV{HOME})
set(TF_DIR                  ${HOME_DIR}/tf/tensorflow)
set(FLATBUFFER_DIR          ${TF_DIR}/tensorflow/contrib/lite/downloads/flatbuffers/include)
set(TFLITE_LIB_DIR          ${HOME_DIR}/android-app/jniLibs/arm64-v8a)
...

include_directories(
    <other include dirs>,
    ${FLATBUFFER_DIR}
    ${TF_DIR}
)

link_directories(${TFLITE_LIB_DIR})
find_library(libtflite tflite)
...

target_link_libraries (
    <other libraries>,
    tflite
)

Coding C++ with Tensorflow lite

Follow the interface code https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/lite/interpreter.h

From Google’s documentation:

// An interpreter for a graph of nodes that input and output from tensors.
// Each node of the graph processes a set of input tensors and produces a
// set of output Tensors. All inputs/output tensors are referenced by index.
//
// Usage:
//
// -- Create basic model
// Interpreter foo(2, 1);
// foo.SetTensorParametersReadWrite(0, ...);
// foo.SetTensorParametersReadOnly(1, ...);
// foo.SetNodeParameters(0, ...)
//
// -- Resize input array to 1 length.
// foo.ResizeInputTensor(0, 1);
// foo.AllocateTensors();
// -- Install array data
// foo.typed_tensor<float>(0)[0] = 3;
// foo.Invoke();
// foo.typed_tensor<float>(0)[0] = 4;
// foo.Invoke();
// -- Resize input array and set data.
// foo.ResizeInputTensor(0, 2);
// foo.AllocateTensors();
// foo.typed_tensor<float>(0)[0] = 4;
// foo.typed_tensor<float>(0)[1] = 8;
// foo.Invoke();
//