cpvrlab/libmediapipe

C API for Google's MediaPipe framework

libmediapipe

About

libmediapipe is a C wrapper around Google's MediaPipe framework. This repository contains scripts that clone MediaPipe, copy the libmediapipe source files into the workspace and build it with Bazel. This produces a shared library that can subsequently be used in CMake/Visual Studio/Xcode/etc. projects without ever touching Bazel again.

Building

Linux

Requirements

• Python 3: apt install python3
• Bazel 5.2.0: https://bazel.build/install/ubuntu
• Numpy: pip3 install numpy

Build and install OpenCV 4.7.0

wget -O opencv.zip https://github.com/opencv/opencv/archive/4.7.0.zip
wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.7.0.zip
unzip opencv.zip
unzip opencv_contrib.zip
mkdir -p opencv_build && cd opencv_build
cmake -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.7.0/modules ../opencv-4.7.0
cmake --build .
cmake --install .

Run the build script

cd <path to libmediapipe>
./build-x86_64-linux.sh --version v0.8.11 --config release --opencv_dir /usr/local

Windows

Requirements

• Python
• Bazel
• Clang
• Visual Studio
• Powershell
• Git with Bash
• Numpy: pip install numpy

Install OpenCV 4.7.0

The installer can be downloaded from https://github.com/opencv/opencv/releases/download/4.7.0/opencv-4.7.0-windows.exe

Run the build script

./build-x86_64-windows.ps1 --version v0.8.11 --config release --opencv_dir <path to opencv build>

MacOS

Requirements

• Python: brew install python
• Bazelisk: brew install bazelisk
• Xcode
• Numpy: pip install numpy or pip3 install numpy

Build and install OpenCV 4.7.0

curl -L -o opencv.zip https://github.com/opencv/opencv/archive/4.7.0.zip
curl -L -o opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.7.0.zip
unzip opencv.zip
unzip opencv_contrib.zip
mkdir -p opencv_build && cd opencv_build
cmake -DBUILD_ZLIB=false -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.7.0/modules ../opencv-4.7.0
cmake --build . -j6
cmake --install .

Run the build script

Architectures

• x86_64: x86_64/AMD64 processors
• aarch64: AArch64/ARM64/Apple silicon processors

cd <path to libmediapipe>
./build-<arch>-macos.sh --version v0.8.11 --config release --opencv_dir /usr/local

Android

Requirements

• Python 3: apt install python3
• Bazel 5.2.0: https://bazel.build/install/ubuntu
• Numpy: pip3 install numpy
• Android SDK
• Android NDK 21

Environment variables

export ANDROID_HOME=<path to android sdk>
export ANDROID_NDK_HOME=<path to android ndk>

Run the build script

cd <path to libmediapipe>
./build-aarch64-android.sh --version v0.8.11 --config release

Usage

The build script generates an output directory, which contains the build directory (libmediapipe-<version>-<arch>-<os>) and the data directory (protobuf files and TensorFlow models). To use the library, add include to your include path and link with libmediapipe.so/mediapipe.lib/libmediapipe.dylib. See example/example.cpp for an example of how to use the library.

Running the Example

Linux & MacOS

Requirements

• CMake

cd <path to libmediapipe>/example
mkdir build && cd build
cmake -DMediaPipe_DIR=<libmediapipe build dir> -DCMAKE_BUILD_TYPE=Release ..
cmake --build .
chmod +x ./example
./example <libmediapipe data dir>

Windows

Requirements

• CMake

cd <path to libmediapipe>/example
mkdir build; cd build
cmake -DMediaPipe_DIR=<libmediapipe build dir> -DOpenCV_DIR=<opencv build dir> ..
cmake --build . --config Release
cd Release
./example.exe <libmediapipe data dir>