-
Port dlib to Android platform
-
You can build it to dynamic or static library for Android. You can also build dlib's sample to Android executable file.
-
You can refer to dlib-android-app which demonstrates dlib-android features
$ git clone --recursive https://github.com/tzutalin/dlib-android.git
$ cd dlib-android
$ ./envsetup
-
Download Android-NDK from Android website.
After downloading, go to the directory to which you downloaded the package to extract it
Export ANDROID_NDK_HOME in ~/.bashrc
$ vim ~/.bashrc
export ANDROID_NDK_HOME=[NDK_PATH]/android-ndk-[version]
export PATH=$PATH:$ANDROID_NDK_HOME
-
Install Android Debug Bride (ADB). You can download it via Android SDK Manager or $ sudo apt-get install android-tools-adb
-
Prepare an Android device for test
-
You can change the compiler architecture in dlib-android/jni/Application.mk
-
The way to build the shared library for Android application
$ cd [dlib-android]
$ python build.py
Alternative way to build native code and copy to the Android Studio's project manually:
$ cd [dlib-android]
$ ndk-build -j 2
$ cp -r libs/* androidstudio-examples/dlib-android-app/dlib/src/main/jniLibs
- Open Android Studio's projects in androidstudio-examples/dlib-android-app to run face detection, face landmark, and so on
├── data # Test data or the models for detection and landmarks
├── dlib # Source files of dlib. It is a submodule
├── jni # Source files of JNI codes and their make files
├── androidstudio-examples # Android Studio's projects use the shared library built by this repo
├── tools # Tools and utilities
├── third_party # Like OpenCV and [miniglog](https://github.com/tzutalin/miniglog)
├── CMakeLists.txt # Use CMake to build instead of using Android.mk
├── LICENSE
└── README.md
- If you have any improvement or you've found any bug, send a pull request with the code.
- Give me a star on this repository
- Add more examples to dlib-android-app