Flower Crown Eevee

Use Kinect on Firefly-RK3399

2017-05-20T05:09:46.000Z

Preparation

Install libfreenect

Note:
libfreenect for Kinect Xbox 360
libfreenect2 for Kinect Xbox One

# download latest code, instead of latest release v0.5.5
$ git clone https://github.com/OpenKinect/libfreenect.git
$ cd libfreenect/
# fix issue with OpenCV, in line 173
$ vi OpenNI2-FreenectDriver/src/DepthStream.hpp
        case XN_STREAM_PROPERTY_ZERO_PLANE_DISTANCE:    // unsigned long long or unsigned int (for OpenNI2/OpenCV)
          if (*pDataSize != sizeof(unsigned long long) && *pDataSize != sizeof(unsigned int))
          {
            LogError("Unexpected size for XN_STREAM_PROPERTY_ZERO_PLANE_DISTANCE");
            return ONI_STATUS_ERROR;
          } else {
            if (*pDataSize != sizeof(unsigned long long)) {
              *(static_cast(data)) = ZERO_PLANE_DISTANCE_VAL;
            } else {
              *(static_cast(data)) = (unsigned int) ZERO_PLANE_DISTANCE_VAL;
            }
          }
          return ONI_STATUS_OK;
# build
$ mkdir build
$ cd build
$ cmake -DBUILD_OPENNI2_DRIVER=ON ..
$ make
# preview
$ ./bin/freenect-glview

Install OpenNI

# download this, aarch64 platform support
$ git clone -b develop https://github.com/joinAero/OpenNI2.git
$ cd OpenNI2/
# deps
$ sudo apt-get install libudev-dev
# or, all
$ sudo apt-get install \
    g++              \
    python           \
    libusb-1.0-0-dev \
    libudev-dev      \
    openjdk-6-jdk    \
    freeglut3-dev    \
    doxygen          \
    graphviz
# build
$ make
# setup
$ echo "export OPENNI2_INCLUDE=`pwd`/Include" >> ~/.bashrc
$ echo "export OPENNI2_REDIST=`pwd`/Bin/AArch64-Release" >> ~/.bashrc

Then, copy the freenect driver to OpenNI2 driver repository:

$ cd libfreenect/build/
$ cd lib/OpenNI2-FreenectDriver/
$ export FREENECT_DRIVER_LIB=libFreenectDriver.so
$ export FREENECT_DRIVER_REAL_LIB=$(python - <<'EOF'
import os
lib = os.environ['FREENECT_DRIVER_LIB']
while os.path.islink(lib):
    lib = os.readlink(lib)
print(lib)
EOF
)
$ ln -sf `pwd`/$FREENECT_DRIVER_REAL_LIB $OPENNI2_REDIST/OpenNI2/Drivers/$FREENECT_DRIVER_LIB

Build OpenCV with OpenNI

NOTE: Please see how to Build OpenCV on Firefly-RK3399.

$ cd opencv/
$ mkdir build
$ cd build/
$ export PY_NAME=$(python -c 'from sys import version_info as v; print("python%d.%d" % v[:2])')
$ export PY_NUMPY_DIR=$(python -c 'import os.path, numpy.core; print(os.path.dirname(numpy.core.__file__))')
$ cmake -DCMAKE_BUILD_TYPE=RELEASE \
-DCMAKE_INSTALL_PREFIX=/usr/local \
\
-DPYTHON2_EXECUTABLE=$(which python) \
-DPYTHON_INCLUDE_DIR=/usr/include/$PY_NAME \
-DPYTHON_INCLUDE_DIR2=/usr/include/aarch64-linux-gnu/$PY_NAME \
-DPYTHON_LIBRARY=/usr/lib/aarch64-linux-gnu/lib$PY_NAME.so \
-DPYTHON2_NUMPY_INCLUDE_DIRS=/usr/lib/$PY_NAME/dist-packages/numpy/core/include/ \
\
-DWITH_OPENNI2=ON \
\
-DBUILD_DOCS=OFF \
-DBUILD_EXAMPLES=OFF \
-DBUILD_TESTS=OFF \
-DBUILD_PERF_TESTS=OFF \
\
-DOPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules \
..
$ make -j$(nproc --all)
$ sudo make install

Usage

C++

kinect.cc:

#include "camera.hpp"
int main(int argc, char const *argv[]) {
    Camera cam(cv::CAP_OPENNI2);
    if (!cam.IsOpened()) {
        std::cerr << "ERROR: Open camera failed" << std::endl;
        return 1;
    }
    std::cout << "\033[1;32mPress ESC/Q to terminate\033[0m\n\n";
    double min, max;
    cv::Mat adjmap, colormap;
    cam.Capture([&](const cv::Mat& frame, const cv::Mat& depthmap) {
        cv::minMaxIdx(depthmap, &min, &max);
        const float scale = 255 / (max-min);
        depthmap.convertTo(adjmap, CV_8UC1, scale, -min*scale);
        //applyColorMap(adjmap, colormap, cv::COLORMAP_AUTUMN);
        cam.DrawInfo(frame);
        cam.DrawInfo(adjmap);
        cv::imshow("frame", frame);
        cv::imshow("depthmap", adjmap);
        const int key = cv::waitKey(10);
        return !(key == 27 || key == 'q' || key == 'Q'); // ESC/Q
    });
    return 0;
}

./build/bin/kinect:

Could get complete C++ code here, Python code here.

NOTE: Unable to retrieve correct image using Python code now. Please see this issue.

References

Using Kinect and other OpenNI compatible depth sensors
OpenNI2 with OpenCV 3 (macOS)
- How to use Kinect using OpenNI2 with OpenCV 3 on macOS

Issues

Failed to use libfreenect form `sudo apt-get install freenect`

Install:

$ sudo apt-get install freenect
$ apt-cache show freenect
$ dpkg -L freenect
$ dpkg -L libfreenect-bin
$ dpkg -L libfreenect-dev
$ dpkg -L libfreenect0.5
$ dpkg -L libfreenect-doc

Cause: not install freenect driver.

Failed to build libfreenect on Firefly-RK3399 (aarch64)

Error when build libfreenect:

[ 94%] Building CXX object OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/src/ColorStream.cpp.o
In file included from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/extern/OpenNI-Linux-x64-2.2.0.33/Include/Driver/OniDriverAPI.h:24:0,
                 from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.hpp:6,
                 from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.cpp:2:
/home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/extern/OpenNI-Linux-x64-2.2.0.33/Include/OniPlatform.h:49:3: error: #error Xiron Platform Abstraction Layer - Unsupported Platform!
 # error Xiron Platform Abstraction Layer - Unsupported Platform!
   ^
In file included from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/extern/OpenNI-Linux-x64-2.2.0.33/Include/Driver/OniDriverAPI.h:25:0,
                 from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.hpp:6,
                 from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.cpp:2:
/home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/extern/OpenNI-Linux-x64-2.2.0.33/Include/OniCTypes.h:115:32: error: expected ')' before '*' token
 typedef void (ONI_CALLBACK_TYPE* OniNewFrameCallback)(OniStreamHandle stream, void* pCookie);
                                ^
...
In file included from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.hpp:7:0,
                 from /home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/ColorStream.cpp:2:
/home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/VideoStream.hpp: In constructor 'FreenectDriver::VideoStream::VideoStream(Freenect::FreenectDevice*)':
/home/firefly/Workspace/libfreenect-0.5.5/OpenNI2-FreenectDriver/src/VideoStream.hpp:34:46: error: 'memset' was not declared in this scope
         memset(&cropping, 0, sizeof(cropping));
                                              ^
OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/build.make:62: recipe for target 'OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/src/ColorStream.cpp.o' failed
make[2]: *** [OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/src/ColorStream.cpp.o] Error 1
CMakeFiles/Makefile2:902: recipe for target 'OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/all' failed
make[1]: *** [OpenNI2-FreenectDriver/CMakeFiles/FreenectDriver.dir/all] Error 2
Makefile:127: recipe for target 'all' failed
make: *** [all] Error 2

Case: not support aarch64 platform if build from latest release v0.5.5.

Solution: use latest code instead of latest release v0.5.5.

See this commit, OpenNI2: Add aarch64 platform detection.

Failed to use OpenNI2 from `sudo apt-get install libopenni2-dev`

Install:

# install
$ sudo apt-get install libopenni2-dev
$ apt-cache show libopenni2-dev
$ dpkg -L libopenni2-dev
# copy driver
$ cd libfreenect/build/
$ cd lib/OpenNI2-FreenectDriver/
$ export FREENECT_DRIVER_LIB=libFreenectDriver.so
$ export FREENECT_DRIVER_REAL_LIB=$(python - <<'EOF'
import os
lib = os.environ['FREENECT_DRIVER_LIB']
while os.path.islink(lib):
    lib = os.readlink(lib)
print(lib)
EOF
)
$ ln -sf `pwd`/$FREENECT_DRIVER_REAL_LIB /usr/lib/OpenNI2/Drivers/$FREENECT_DRIVER_LIB

Error when run sample:

OpenCV Error: Unspecified error (OpenCVKinect: Device open failed see:  DeviceOpen using default: no devices found

) in CvCapture_OpenNI2, file /home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp, line 237
VIDEOIO(cvCreateCameraCapture_OpenNI2(index)): raised OpenCV exception:

/home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp:237: error: (-2) OpenCVKinect: Device open failed see:   DeviceOpen using default: no devices found

 in function CvCapture_OpenNI2

Case: not found freenect driver! why?

Failed to use OpenNI2 from `OpenNI-Linux-Arm-2.2.0.33.tar.bz2`

Install:

# download
$ curl -O https://s3.amazonaws.com/com.occipital.openni/OpenNI-Linux-Arm-2.2.0.33.tar.bz2
$ tar jxf OpenNI-Linux-Arm-2.2.0.33.tar.bz2
$ cd OpenNI-Linux-Arm-2.2/
# install
$ sudo ./install.sh
$ cat OpenNIDevEnvironment >> ~/.bashrc
$ source ~/.bashrc

OpenNI2 Wiki

Then, copy the freenect driver to OpenNI2 driver repository:

$ cd libfreenect/build/
$ cd lib/OpenNI2-FreenectDriver/
$ export FREENECT_DRIVER_LIB=libFreenectDriver.so
$ export FREENECT_DRIVER_REAL_LIB=$(python - <<'EOF'
import os
lib = os.environ['FREENECT_DRIVER_LIB']
while os.path.islink(lib):
    lib = os.readlink(lib)
print(lib)
EOF
)
$ ln -sf `pwd`/$FREENECT_DRIVER_REAL_LIB $OPENNI2_REDIST/OpenNI2/Drivers/$FREENECT_DRIVER_LIB

Error when rebuild OpenCV:

[ 61%] Building CXX object modules/videoio/CMakeFiles/opencv_videoio.dir/src/cap_openni2.cpp.o
In file included from /home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Include/OpenNI.h:24:0,
                 from /home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp:69:
/home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Include/OniPlatform.h:49:3: error: #error Xiron Platform Abstraction Layer - Unsupported Platform!
 # error Xiron Platform Abstraction Layer - Unsupported Platform!
   ^
In file included from /home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Include/OniCAPI.h:25:0,
                 from /home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Include/OpenNI.h:28,
                 from /home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp:69:
/home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Include/OniCTypes.h:115:32: error: expected ')' before '*' token
 typedef void (ONI_CALLBACK_TYPE* OniNewFrameCallback)(OniStreamHandle stream, void* pCookie);
                                ^
...
/home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp: In member function 'bool CvCapture_OpenNI2::setDepthGeneratorProperty(int, double)':
/home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp:612:60: warning: logical not is only applied to the left hand side of comparison [-Wlogical-not-parentheses]
                     if( !device.getImageRegistrationMode() == mode )
                                                            ^
modules/videoio/CMakeFiles/opencv_videoio.dir/build.make:206: recipe for target 'modules/videoio/CMakeFiles/opencv_videoio.dir/src/cap_openni2.cpp.o' failed
make[2]: *** [modules/videoio/CMakeFiles/opencv_videoio.dir/src/cap_openni2.cpp.o] Error 1
CMakeFiles/Makefile2:3987: recipe for target 'modules/videoio/CMakeFiles/opencv_videoio.dir/all' failed
make[1]: *** [modules/videoio/CMakeFiles/opencv_videoio.dir/all] Error 2

Case: not support aarch64 platform.

Solution:

$ cd OpenNI-Linux-Arm-2.2
# add `__aarch64__` in line 42
$ vi Include/OniPlatform.h
#elif (__linux__ && (__arm__ || __aarch64__))

Error when rebuild OpenCV again:

[ 60%] Linking CXX shared library ../../lib/libopencv_videoio.so
/usr/bin/ld: skipping incompatible /home/firefly/Workspace/OpenNI-Linux-Arm-2.2/Redist/libOpenNI2.so when searching for -lOpenNI2
/usr/bin/ld: cannot find -lOpenNI2
collect2: error: ld returned 1 exit status
modules/videoio/CMakeFiles/opencv_videoio.dir/build.make:281: recipe for target 'lib/libopencv_videoio.so.3.2.0' failed
make[2]: *** [lib/libopencv_videoio.so.3.2.0] Error 1
CMakeFiles/Makefile2:3987: recipe for target 'modules/videoio/CMakeFiles/opencv_videoio.dir/all' failed
make[1]: *** [modules/videoio/CMakeFiles/opencv_videoio.dir/all] Error 2
Makefile:149: recipe for target 'all' failed
make: *** [all] Error 2

Case: not incompatible “libOpenNI2.so”.

Solution: build form source.

Failed to use libfreenect with OpenNI2 (from source)

Error when run sample:

OpenNI2-FreenectDriver: (ERROR) Unexpected size for XN_STREAM_PROPERTY_ZERO_PLANE_DISTANCE
OpenCV Error: Unspecified error (CvCapture_OpenNI2::readCamerasParams : Could not read virtual plane distance!
) in readCamerasParams, file /home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp, line 404
write_register: 0x0005 <= 0x00
write_register: 0x0006 <= 0x00
OpenNI2-FreenectDriver: Closing device freenect://0
VIDEOIO(cvCreateCameraCapture_OpenNI2(index)): raised OpenCV exception:

/home/firefly/Workspace/opencv/modules/videoio/src/cap_openni2.cpp:404: error: (-2) CvCapture_OpenNI2::readCamerasParams : Could not read virtual plane distance!
 in function readCamerasParams

Solution: fix this issue, plz see here.

Build OpenCV on Firefly-RK3399

2017-05-17T04:49:56.000Z

Install

Required Packages

# compiler ✓
$ sudo apt-get install build-essential
# required ✓
$ sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
# optional ✓
$ sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev

Getting OpenCV Source Code

1
2
3

$ git clone https://github.com/opencv/opencv.git
$ cd opencv/
$ git checkout 3.2.0

If use extra modules:

1
2
3

$ git clone https://github.com/opencv/opencv_contrib.git
$ cd opencv_contrib/
$ git checkout 3.2.0

Building OpenCV from Source

$ cd opencv/
$ mkdir build
$ cd build/
$ export PY_NAME=$(python -c 'from sys import version_info as v; print("python%d.%d" % v[:2])')
$ export PY_NUMPY_DIR=$(python -c 'import os.path, numpy.core; print(os.path.dirname(numpy.core.__file__))')
$ cmake -DCMAKE_BUILD_TYPE=RELEASE \
-DCMAKE_INSTALL_PREFIX=/usr/local \
\
-DPYTHON2_EXECUTABLE=$(which python) \
-DPYTHON_INCLUDE_DIR=/usr/include/$PY_NAME \
-DPYTHON_INCLUDE_DIR2=/usr/include/aarch64-linux-gnu/$PY_NAME \
-DPYTHON_LIBRARY=/usr/lib/aarch64-linux-gnu/lib$PY_NAME.so \
-DPYTHON2_NUMPY_INCLUDE_DIRS=/usr/lib/$PY_NAME/dist-packages/numpy/core/include/ \
\
-DBUILD_DOCS=OFF \
-DBUILD_EXAMPLES=OFF \
-DBUILD_TESTS=OFF \
-DBUILD_PERF_TESTS=OFF \
\
-DOPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules \
..
$ make -j$(nproc --all)
$ sudo make install

cmake result:

-- General configuration for OpenCV 3.2.0 =====================================
--   Version control:               3.2.0
--
--   Extra modules:
--     Location (extra):            /home/firefly/Workspace/opencv_contrib/modules
--     Version control (extra):     3.2.0
--
--   Platform:
--     Timestamp:                   2017-05-13T13:43:22Z
--     Host:                        Linux 4.4.16 aarch64
--     CMake:                       3.5.1
--     CMake generator:             Unix Makefiles
--     CMake build tool:            /usr/bin/make
--     Configuration:               RELEASE
--
--   C/C++:
--     Built as dynamic libs?:      YES
--     C++ Compiler:                /usr/bin/c++  (ver 5.4.0)
--     C++ flags (Release):         -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wno-narrowing -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fvisibility=hidden -fvisibility-inlines-hidden -O3 -DNDEBUG  -DNDEBUG
--     C++ flags (Debug):           -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wno-narrowing -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fvisibility=hidden -fvisibility-inlines-hidden -g  -O0 -DDEBUG -D_DEBUG
--     C Compiler:                  /usr/bin/cc
--     C flags (Release):           -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wno-narrowing -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fvisibility=hidden -O3 -DNDEBUG  -DNDEBUG
--     C flags (Debug):             -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wno-narrowing -Wno-comment -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fvisibility=hidden -g  -O0 -DDEBUG -D_DEBUG
--     Linker flags (Release):
--     Linker flags (Debug):
--     ccache:                      NO
--     Precompiled headers:         YES
--     Extra dependencies:          /usr/lib/aarch64-linux-gnu/libpng.so /usr/lib/aarch64-linux-gnu/libz.so /usr/lib/aarch64-linux-gnu/libtiff.so /usr/lib/aarch64-linux-gnu/libjasper.so /usr/lib/aarch64-linux-gnu/libjpeg.so gtk-x11-2.0 gdk-x11-2.0 pangocairo-1.0 atk-1.0 cairo gdk_pixbuf-2.0 gio-2.0 pangoft2-1.0 pango-1.0 gobject-2.0 glib-2.0 fontconfig freetype gthread-2.0 dc1394 avcodec-ffmpeg avformat-ffmpeg avutil-ffmpeg swscale-ffmpeg dl m pthread rt
--     3rdparty dependencies:       libwebp IlmImf tegra_hal
--
--   OpenCV modules:
--     To be built:                 core flann imgproc ml photo reg surface_matching video freetype fuzzy imgcodecs shape videoio highgui objdetect plot superres xobjdetect xphoto bgsegm bioinspired dpm face features2d line_descriptor saliency text calib3d ccalib datasets rgbd stereo videostab xfeatures2d ximgproc aruco optflow phase_unwrapping stitching structured_light python2
--     Disabled:                    world contrib_world
--     Disabled by dependency:      tracking
--     Unavailable:                 cudaarithm cudabgsegm cudacodec cudafeatures2d cudafilters cudaimgproc cudalegacy cudaobjdetect cudaoptflow cudastereo cudawarping cudev java python3 ts viz cnn_3dobj cvv dnn hdf matlab sfm
--
--   GUI:
--     QT:                          NO
--     GTK+ 2.x:                    YES (ver 2.24.30)
--     GThread :                    YES (ver 2.48.2)
--     GtkGlExt:                    NO
--     OpenGL support:              NO
--     VTK support:                 NO
--
--   Media I/O:
--     ZLib:                        /usr/lib/aarch64-linux-gnu/libz.so (ver 1.2.8)
--     JPEG:                        /usr/lib/aarch64-linux-gnu/libjpeg.so (ver )
--     WEBP:                        build (ver 0.3.1)
--     PNG:                         /usr/lib/aarch64-linux-gnu/libpng.so (ver 1.2.54)
--     TIFF:                        /usr/lib/aarch64-linux-gnu/libtiff.so (ver 42 - 4.0.6)
--     JPEG 2000:                   /usr/lib/aarch64-linux-gnu/libjasper.so (ver 1.900.1)
--     OpenEXR:                     build (ver 1.7.1)
--     GDAL:                        NO
--     GDCM:                        NO
--
--   Video I/O:
--     DC1394 1.x:                  NO
--     DC1394 2.x:                  YES (ver 2.2.4)
--     FFMPEG:                      YES
--       avcodec:                   YES (ver 56.60.100)
--       avformat:                  YES (ver 56.40.101)
--       avutil:                    YES (ver 54.31.100)
--       swscale:                   YES (ver 3.1.101)
--       avresample:                NO
--     GStreamer:                   NO
--     OpenNI:                      NO
--     OpenNI PrimeSensor Modules:  NO
--     OpenNI2:                     NO
--     PvAPI:                       NO
--     GigEVisionSDK:               NO
--     Aravis SDK:                  NO
--     UniCap:                      NO
--     UniCap ucil:                 NO
--     V4L/V4L2:                    NO/YES
--     XIMEA:                       NO
--     Xine:                        NO
--     gPhoto2:                     NO
--
--   Parallel framework:            pthreads
--
--   Other third-party libraries:
--     Use IPP:                     NO
--     Use VA:                      NO
--     Use Intel VA-API/OpenCL:     NO
--     Use Lapack:                  NO
--     Use Eigen:                   NO
--     Use Cuda:                    NO
--     Use OpenCL:                  YES
--     Use OpenVX:                  NO
--     Use custom HAL:              YES (carotene (ver 0.0.1))
--
--   OpenCL:                        
--     Include path:                /home/firefly/Workspace/opencv/3rdparty/include/opencl/1.2
--     Use AMDFFT:                  NO
--     Use AMDBLAS:                 NO
--
--   Python 2:
--     Interpreter:                 /usr/bin/python (ver 2.7.12)
--     Libraries:                   /usr/lib/aarch64-linux-gnu/libpython2.7.so (ver 2.7.12)
--     numpy:                       /usr/lib/python2.7/dist-packages/numpy/core/include/ (ver 1.11.0)
--     packages path:               lib/python2.7/dist-packages
--
--   Python 3:
--     Interpreter:                 /usr/bin/python3 (ver 3.5.2)
--
--   Python (for build):            /usr/bin/python
--
--   Java:
--     ant:                         NO
--     JNI:                         NO
--     Java wrappers:               NO
--     Java tests:                  NO
--
--   Matlab:                        Matlab not found or implicitly disabled
--
--   Tests and samples:
--     Tests:                       NO
--     Performance tests:           NO
--     C/C++ Examples:              NO
--
--   Install path:                  /usr/local
--
--   cvconfig.h is in:              /home/firefly/Workspace/opencv/build
-- -----------------------------------------------------------------

Extra Links

Installation in Linux

Examples

Canny Edge Detection

canny.py:

#!/usr/bin/env python
import cv2
import numpy as np
def main():
    img = cv2.imread('../data/messi5.jpg', 0)
    edges = cv2.Canny(img, 100, 200)
    cv2.imshow('Original & Edge', np.vstack((img, edges)))
    cv2.waitKey()
    cv2.destroyAllWindows()
if __name__ == '__main__':
    main()

Appendix

Install pip

$ python --version
Python 2.7.12
$ curl -O https://bootstrap.pypa.io/get-pip.py
$ python get-pip.py
$ vi ~/.pip/pip.conf
[global]
format=columns
index-url=http://mirrors.aliyun.com/pypi/simple/
[install]
trusted-host=mirrors.aliyun.com

Headless Firefly-RK3399 Setup

2017-05-10T13:40:10.000Z

Firefly-RK3399

Flash Image

Windows:

Download here, 或这里
- Ubuntu 16.04 image: Firefly-RK3399_Ubuntu16.04_201703181519.img
- RK Tool: AndroidTool_Release_v2.38.rar
- RK USB driver: DriverAssitant_v4.5.rar
Install RK USB Driver
- Run “DriverAssitant_v4.5/DriverInstall.exe”
Connect device to host
1. Connect the power adapter to board.
2. Use Type-C cable to connect host and device together.
3. Press and hold RECOVERY key.
4. Shortly press RESET key.
5. After around two seconds, release RECOVERY key.
  
  The host will prompt to have new device detected and configured. Open the Device Management, you’ll find a new device name “Rockusb Device”, as shown below. Return to previous step to reinstall driver if it is not shown.
Flush image
1. Run “AndroidTool_Release_v2.38/AndroidTool.exe” as Administrator.
  - Edit “config.ini”, set “Selected=2” to English
2. Switch to “Upgrade Firmware” tab page.
3. Click “Firmware” button and open the image file.
  - Detail information of the image file, like version and chip, is shown.
4. Click “Upgrade” button to start flash.
5. If upgrade fails, please try “LowerFormat” in the “Download Image” tab page first, then try again.
  
  WARN: If you flash firmware laoder different version of the original machine, please click “Erase Flash” before upgrading the firmware.

Setup Firefly

Connect Display or TV with HDMI cable
Connect the power adapter to the board
- The board will boot automatically once power on
Plug USB mouse & keyboard to the board

Now, you can control the board.

Username: firefly
Password: firefly

Connect the Wi-Fi
Enable SSH server
- Detect SSH server, ps aux | grep sshd
  - Otherwise, sudo apt-get install openssh-server and sudo /etc/init.d/ssh start
- Get ip address, ifconfig | grep "inet "
- SSH from remote pc, ssh firefly@
Screen Sharing from Firefly
- Install, sudo apt-get install vino
- Config, vino-preferences
- Start, /usr/lib/vino/vino-server –sm-disable`
- Autostart, open “Session and Startup” and check “Desktop Sharing”,
Remote access from Ubuntu
- Open “Remmina Remote Desktop Client” and add connection,
- Or, remote access from Mac, plz see here.

Finally, you can control the board from remote PC, without extra accessories or devices.

NOTE: You could get ip address of the board from your router’s home page.

If you wanna access the Firefly’s files, you can do as follows:

Install Samba on Firefly,

# samba
sudo apt-get install samba samba-common-bin
# config samba
sudo vi /etc/samba/smb.conf
[homes]
  read only = no
# restart
sudo /etc/init.d/samba restart
# add user
sudo smbpasswd -a firefly

Open “Finder” on Mac,
- “Go > Connect to Server”
- Then, select “firefly” volume

If you wanna setup other WiFi connections, you can do as follows:

Add a WiFi connection,
- Run nmtui, then “Edit a connection > Add > Wi-Fi”
- Field “Device” is “wlan0 mac address”, run ip addr to get
- Enter “OK”, done

Connection file is created here,

1	$ ll /etc/NetworkManager/system-connections/

Connect a WiFi with SSID found nearby,

1	$ nmcli device wifi con "ssid" password "password"

Screen Sharing from Ubuntu to Mac

2017-05-09T09:00:45.000Z

Ubuntu

1) Open “Desktop Sharing” and setup,

2) sudo apt-get install dconf-tools and run dconf-editor,

Or,

1 2	$ export DISPLAY=:0 $ dconf write /org/gnome/desktop/remote-access/require-encryption false

Mac

1) Spotlight Search “Screen Sharing” and connect to,

2) View and control “Ubuntu Desktop”,

References

Headless Raspberry Pi Setup

2017-05-09T01:12:33.000Z

Install

Download “RASPBIAN JESSIE LITE” without PIXEL desktop.
Unzip it to get the image file (.img) for writing to your SD card.
Setup headless settings, if don’t have an extra keyboard or HDMI cable.
1. Add “ssh” file to the SD card root. Then boot your Pi.
2. Connect it to the router with wire. Then get its ip from the router.
3. Use SSH to connect to your Pi.
  - ssh pi@, password: raspberry.
4. Configure your Pi.
  - sudo raspi-config
5. Setting WIFI up and reboot wireless.
  - sudo vi /etc/wpa_supplicant/wpa_supplicant.conf
    1
    2
    3
    4
    5
    6
    network={
    ssid="ssid"
    psk="password"
    priority=1
    id_str="home"
    }
  - sudo reboot
6. Setting static ip.
  - ifconfig
  - sudo nano /etc/dhcpcd.conf
    1
    2
    3
    4
    # Custom static IP address
    interface wlan0
    static ip_address = 192.168.1.121/24
    static routers = 192.168.1.9
Writing the image to the SD card, see the guide for your system:
- Linux
- Mac OS
- Windows

Prepare

1	sudo apt-get update

Vim

# vim
sudo apt-get remove vim-common
sudo apt-get install vim
# config vim
sudo vi /etc/vim/vimrc
# 显示行号
set nu
# tab 退四格
set tabstop=4
# 语法高亮
syntax on
# 显示空白符
# http://stackoverflow.com/questions/1675688/make-vim-show-all-white-spaces-as-a-character
# howto: http://vimsheet.com/

Samba

# samba
sudo apt-get install samba samba-common-bin
# config samba
sudo vi /etc/samba/smb.conf
[homes]
  read only = no
# restart
sudo /etc/init.d/samba restart
# add user
sudo smbpasswd -a pi
# Connect it under "Shared of Finder on Mac"

Python

python --version
# pip
which pip
curl -O https://bootstrap.pypa.io/get-pip.py
sudo python get-pip.py
# config pip
vi ~/.pip/pip.conf
[global]
format=columns
index-url = https://mirrors.aliyun.com/pypi/simple/
trusted-host = mirrors.aliyun.com
pip list [--outdated]
sudo pip install --upgrade (package)
# GPIO
sudo pip install RPi.GPIO

Then

1	sudo apt-get remove

Usage

Light a LED and so on.

RPi.GPIO: https://sourceforge.net/p/raspberry-gpio-python/wiki/Home/

vi led.py:

#!/usr/bin/env python
# -*- coding:utf-8 -*-
import RPi.GPIO as GPIO
import time
print(GPIO.RPI_INFO)
LED = 26
GPIO.setmode(GPIO.BCM)
GPIO.setup(LED, GPIO.OUT)
try:
    while True:
        GPIO.output(LED, GPIO.HIGH)
        time.sleep(1)
        GPIO.output(LED, GPIO.LOW)
        time.sleep(1)
except:
    print("except")
GPIO.cleanup()

vi key.py:

#!/usr/bin/env python
# -*- coding:utf-8 -*-
import RPi.GPIO as GPIO
import time
KEY = 20
GPIO.setmode(GPIO.BCM)
GPIO.setup(KEY, GPIO.IN, GPIO.PUD_UP)
while True:
    time.sleep(0.05)
    if GPIO.input(KEY) == 0:
        print("KEY PRESS")
        while GPIO.input(KEY) == 0:
            time.sleep(0.01)

vi event.py:

#!/usr/bin/env python
# -*- coding:utf-8 -*-
import RPi.GPIO as GPIO
import time
KEY = 20
def MyInterrupt(KEY):
    print("KEY PRESS")
GPIO.setmode(GPIO.BCM)
GPIO.setup(KEY, GPIO.IN, GPIO.PUD_UP)
GPIO.add_event_detect(KEY, GPIO.FALLING, MyInterrupt, 200)
while True:
    time.sleep(1)

Build OpenCV for Android with CUDA

2017-05-05T07:31:13.000Z

Prerequisites

Ubuntu 14.04 LTS
- Download here
NVIDIA CodeWorks for Android 1R4
- Doc here
- Download here
- Could only select actions: CUDA ToolKit 7.0, CUDA for Android 7.0
Android Studio
- Add C and C++ Code to Your Project

Prepare

# Set environment variables
$ vi ~/.bashrc
# Android SDK & NDK
export ANDROID_SDK=$HOME/Android/Sdk
export ANDROID_NDK=$ANDROID_SDK/ndk-bundle
export ANDROID_STANDALONE_TOOLCHAIN=$ANDROID_NDK/toolchains
export PATH=$PATH:$ANDROID_SDK/tools:$ANDROID_SDK/tools/bin:$ANDROID_SDK/platform-tools:$ANDROID_NDK
# Java SDK 1.8 (could ignore, for latest android command-line tool)
export JAVA_HOME=$HOME/Develop/jdk1.8.0_131
export PATH=$JAVA_HOME/bin:$PATH
$ source ~/.bashrc
# Enable OpenCV Java wrappers
$ sudo apt-get install ant
# Downgrade android command-line tool
$ cd $ANDROID_SDK
$ mv tools/ tools-backup/
$ curl -O https://dl.google.com/android/repository/tools_r25.2.3-linux.zip
$ unzip -q tools_r25.2.3-linux.zip

Build

$ cd opencv-3.2.0/platforms/
$ mkdir -p build_android_arm
$ cd build_android_arm/
$ cmake -DCMAKE_BUILD_TYPE=RELEASE \
-DCMAKE_BUILD_WITH_INSTALL_RPATH=ON \
-DCMAKE_TOOLCHAIN_FILE=../android/android.toolchain.cmake \
\
-DCUDA_TOOLKIT_ROOT_DIR=$HOME/NVPACK/cuda-7.0 \
-DCUDA_ARCH_BIN="5.3" \
-DCUDA_ARCH_PTX="" \
-DCUDA_FAST_MATH=ON \
-DWITH_CUDA=ON \
-DWITH_CUFFT=OFF \
-DWITH_CUBLAS=OFF \
-DWITH_OPENCL=OFF \
\
-DBUILD_SHARED_LIBS=OFF \
-DBUILD_DOCS=OFF \
-DBUILD_EXAMPLES=OFF \
-DBUILD_TESTS=OFF \
-DBUILD_PERF_TESTS=OFF \
../..

Result:

-- General configuration for OpenCV 3.2.0 =====================================
--   Version control:               unknown
--
--   Platform:
--     Timestamp:                   2017-05-05T05:05:05Z
--     Host:                        Linux 4.4.0-75-generic x86_64
--     Target:                      Linux 1 armv7-a
--     CMake:                       2.8.12.2
--     CMake generator:             Unix Makefiles
--     CMake build tool:            /usr/bin/make
--     Configuration:               RELEASE
--
--   C/C++:
--     Built as dynamic libs?:      NO
--     C++ Compiler:                /home/john/Android/Sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/bin/arm-linux-androideabi-g++  (ver 4.9)
--     C++ flags (Release):         -fexceptions -frtti -fpic -Wno-psabi --sysroot=/home/john/Android/Sdk/ndk-bundle/platforms/android-9/arch-arm -funwind-tables -finline-limit=64 -fsigned-char -no-canonical-prefixes -march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16 -fdata-sections -ffunction-sections -Wa,--noexecstack    -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wno-narrowing -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -fomit-frame-pointer -mfp16-format=ieee -fvisibility=hidden -fvisibility-inlines-hidden -mthumb -fomit-frame-pointer -fno-strict-aliasing -O3 -DNDEBUG  -DNDEBUG
--     C++ flags (Debug):           -fexceptions -frtti -fpic -Wno-psabi --sysroot=/home/john/Android/Sdk/ndk-bundle/platforms/android-9/arch-arm -funwind-tables -finline-limit=64 -fsigned-char -no-canonical-prefixes -march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16 -fdata-sections -ffunction-sections -Wa,--noexecstack    -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wno-narrowing -Wno-delete-non-virtual-dtor -Wno-comment -fdiagnostics-show-option -fomit-frame-pointer -mfp16-format=ieee -fvisibility=hidden -fvisibility-inlines-hidden -marm -fno-omit-frame-pointer -fno-strict-aliasing -O0 -g -DDEBUG -D_DEBUG  -O0 -DDEBUG -D_DEBUG
--     C Compiler:                  /home/john/Android/Sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/bin/arm-linux-androideabi-gcc
--     C flags (Release):           -fexceptions -fpic -Wno-psabi --sysroot=/home/john/Android/Sdk/ndk-bundle/platforms/android-9/arch-arm -funwind-tables -finline-limit=64 -fsigned-char -no-canonical-prefixes -march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16 -fdata-sections -ffunction-sections -Wa,--noexecstack    -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wno-narrowing -Wno-comment -fdiagnostics-show-option -fomit-frame-pointer -mfp16-format=ieee -fvisibility=hidden -mthumb -fomit-frame-pointer -fno-strict-aliasing -O3 -DNDEBUG  -DNDEBUG
--     C flags (Debug):             -fexceptions -fpic -Wno-psabi --sysroot=/home/john/Android/Sdk/ndk-bundle/platforms/android-9/arch-arm -funwind-tables -finline-limit=64 -fsigned-char -no-canonical-prefixes -march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16 -fdata-sections -ffunction-sections -Wa,--noexecstack    -fsigned-char -W -Wall -Werror=return-type -Werror=non-virtual-dtor -Werror=address -Werror=sequence-point -Wformat -Werror=format-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wno-narrowing -Wno-comment -fdiagnostics-show-option -fomit-frame-pointer -mfp16-format=ieee -fvisibility=hidden -marm -fno-omit-frame-pointer -fno-strict-aliasing -O0 -g -DDEBUG -D_DEBUG  -O0 -DDEBUG -D_DEBUG
--     Linker flags (Release):      -Wl,--fix-cortex-a8 -Wl,--no-undefined -Wl,-allow-shlib-undefined -Wl,--gc-sections -Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now
--     Linker flags (Debug):        -Wl,--fix-cortex-a8 -Wl,--no-undefined -Wl,-allow-shlib-undefined -Wl,--gc-sections -Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now
--     ccache:                      YES
--     Precompiled headers:         NO
--     Extra dependencies:          z dl m log cudart nppc nppi npps -L/home/john/NVPACK/cuda-7.0/targets/armv7-linux-androideabi/lib -L/home/john/NVPACK/cuda-7.0/targets/armv7-linux-androideabi/lib/stubs
--     3rdparty dependencies:       libjpeg libwebp libpng libtiff libjasper IlmImf tegra_hal
--
--   OpenCV modules:
--     To be built:                 cudev core cudaarithm flann imgproc ml video cudabgsegm cudafilters cudaimgproc cudawarping imgcodecs photo shape videoio cudacodec highgui objdetect features2d calib3d cudafeatures2d cudalegacy cudaobjdetect cudaoptflow cudastereo java stitching superres videostab
--     Disabled:                    world
--     Disabled by dependency:      -
--     Unavailable:                 python2 ts viz
--
--   Android:
--     Android ABI:                 armeabi-v7a
--     STL type:                    gnustl_static
--     Native API level:            android-9
--     SDK target:                  android-21
--     Android NDK:                 /home/john/Android/Sdk/ndk-bundle (toolchain: arm-linux-androideabi-4.9)
--     android tool:                /home/john/Android/Sdk/tools/android (Android SDK Tools, revision 25.2.3.)
--     Google Play manager:         NO
--     Android examples:            YES
--
--   GUI:
--     GTK+:                        NO
--     GThread :                    NO
--     GtkGlExt:                    NO
--     OpenGL support:              NO
--     VTK support:                 NO
--
--   Media I/O:
--     ZLib:                        z (ver 1.2.3)
--     JPEG:                        build (ver 90)
--     WEBP:                        build (ver 0.3.1)
--     PNG:                         build (ver 1.6.24)
--     TIFF:                        build (ver 42 - 4.0.2)
--     JPEG 2000:                   build (ver 1.900.1)
--     OpenEXR:                     build (ver 1.7.1)
--     GDAL:                        NO
--     GDCM:                        NO
--
--   Video I/O:
--
--   Parallel framework:            pthreads
--
--   Other third-party libraries:
--     Use IPP:                     NO
--     Use Eigen:                   YES (ver 3.2.0)
--     Use Cuda:                    YES (ver 7.0)
--     Use OpenCL:                  NO
--     Use OpenVX:                  NO
--     Use custom HAL:              YES (carotene (ver 0.0.1))
--
--   NVIDIA CUDA
--     Use CUFFT:                   NO
--     Use CUBLAS:                  NO
--     USE NVCUVID:                 NO
--     NVIDIA GPU arch:             53
--     NVIDIA PTX archs:
--     Use fast math:               YES
--
--   Python 2:
--     Interpreter:                 /usr/bin/python2.7 (ver 2.7.6)
--
--   Python 3:
--     Interpreter:                 /usr/bin/python3.4 (ver 3.4.3)
--
--   Python (for build):            /usr/bin/python2.7
--
--   Java:
--     ant:                         /usr/bin/ant (ver 1.9.3)
--     Java wrappers:               YES
--     Java tests:                  NO
--
--   Matlab:                        NO
--
--   Tests and samples:
--     Tests:                       NO
--     Performance tests:           NO
--     C/C++ Examples:              NO
--
--   Install path:                  /home/john/Workspace/opencv-3.2.0/platforms/build_android_arm/install
--
--   cvconfig.h is in:              /home/john/Workspace/opencv-3.2.0/platforms/build_android_arm
-- -----------------------------------------------------------------

Make:

1
2
3

$ make -j8
# Or,
$ make -j$(python -c 'import multiprocessing as mp; print(mp.cpu_count())')

Alternatives (ignored):

$ cd opencv-3.2.0/platforms/
$ ./scripts/cmake_android_arm.sh
# Tegra X1: https://developer.nvidia.com/content/tegra-x1
-D CUDA_GENERATION=Maxwell

Usage

Open “Android Studio” and,

“Import Module” from “build_android_arm” directory.
- Exclude “jni”, only keep “aidl”, “java”, “res/values/attrs.xml”

Ensure it’s the same file structure as the offical OpenCV4Android.

About the sample project, you could find here.

Issues

Issue when `cmake`

Failed to get list of installed Android targets.

CMake Warning at cmake/OpenCVDetectAndroidSDK.cmake:205 (message):
  Can not find any SDK target compatible with: 9 11

And,

$ android list target -c
*************************************************************************
The "android" command is deprecated.
For manual SDK, AVD, and project management, please use Android Studio.
For command-line tools, use tools/bin/sdkmanager and tools/bin/avdmanager
*************************************************************************
Running /home/john/Android/Sdk/tools/bin/avdmanager list target -c
Exception in thread "main" java.lang.UnsupportedClassVersionError: com/android/sdklib/tool/AvdManagerCli : Unsupported major.minor version 52.0
    at java.lang.ClassLoader.defineClass1(Native Method)
    at java.lang.ClassLoader.defineClass(ClassLoader.java:803)
    at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142)
    at java.net.URLClassLoader.defineClass(URLClassLoader.java:442)
    at java.net.URLClassLoader.access$100(URLClassLoader.java:64)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:354)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:348)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.net.URLClassLoader.findClass(URLClassLoader.java:347)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:425)
    at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:308)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:358)
    at sun.launcher.LauncherHelper.checkAndLoadMain(LauncherHelper.java:482)

Solution:

Install Java SDK 1.8
- Fix “UnsupportedClassVersionError …”

Amend match of android sdk targets

$ cd opencv-3.2.0/
$ vi cmake/OpenCVDetectAndroidSDK.cmake
# line 98, change "[^\n]+" to "android-[0-9]+"
string(REGEX MATCHALL "android-[0-9]+" ANDROID_SDK_TARGETS "${ANDROID_SDK_TARGETS}")

Or, cmake with -DANDROID_EXECUTABLE=$ANDROID_SDK/tools/bin/avdmanager
WARN： However, make still will fail in “Generating OpenCV Android library project.”

Issue when `make`

[ 92%] Generating OpenCV Android library project. SDK target: android-21
*************************************************************************
The "android" command is deprecated.
For manual SDK, AVD, and project management, please use Android Studio.
For command-line tools, use tools/bin/sdkmanager and tools/bin/avdmanager
*************************************************************************
Invalid or unsupported command "--silent create lib-project --path /home/john/Workspace/opencv-3.2.0/platforms/build_android_arm --target android-21 --name OpenCV --package org.opencv"

Supported commands are:
android list target
android list avd
android list device
android create avd
android move avd
android delete avd
android list sdk
android update sdk
make[2]: *** [build.xml] Error 2
make[1]: *** [modules/java/CMakeFiles/opencv_java.dir/all] Error 2
make: *** [all] Error 2

Solution:

Downgrade android command-line tool

$ cd $ANDROID_SDK
$ mv tools/ tools-backup/
$ curl -O https://dl.google.com/android/repository/tools_r25.2.3-linux.zip
$ unzip -q tools_r25.2.3-linux.zip

Learn more:

Tools 25.3.1 deprecation of ‘android’ command line tool breaks Unity3D build

References

Use ccache, distcc, dmucs to accelerate builds

2017-05-05T03:29:26.000Z

ccache is a compiler cache. It speeds up recompilation by caching previous compilations and detecting when the same compilation is being done again. Supported languages are C, C++, Objective-C and Objective-C++.

Mac:

# install
$ brew install ccache
# where
$ brew ls ccache
$ which ccache
$ brew info ccache
# env
$ vi ~/.bash_profile
    export PATH=/usr/local/opt/ccache/libexec:$PATH
# verify
$ source ~/.bash_profile
$ which clang
/usr/local/opt/ccache/libexec/clang

Ubuntu:

# install
$ sudo apt-get install ccache
# where
$ whereis ccache
$ which ccache
# env
$ vi ~/.bashrc
    export PATH=/usr/lib/ccache:$PATH
# verify
$ source ~/.bashrc
$ which g++ gcc
/usr/lib/ccache/g++
/usr/lib/ccache/gcc

Using CCache on Mac

distcc on Ubuntu

distcc is a program to distribute compilation of C or C++ code across several machines on a network. distcc should always generate the same results as a local compile, is simple to install and use, and is often two or more times faster than a local compile.

Install:

# install
$ sudo apt-get install distcc
# optional
$ sudo apt-get install distccmon-gnome
$ distcc --help
# show ip
$ ifconfig | grep "inet " | grep -v 127.0.0.1
192.168.199.247
# config
$ sudo vi /etc/default/distcc
    STARTDISTCC="true"
    ALLOWEDNETS="192.168.0.0/16"
    LISTENER=""
    ZEROCONF="false"
# start server
$ sudo /etc/init.d/distcc start
# reboot & verify
$ ps aux | grep distccd

Usage:

# download
$ curl -R -O http://www.lua.org/ftp/lua-5.3.4.tar.gz
$ tar zxf lua-5.3.4.tar.gz
$ cd lua-5.3.4
# build
$ time make linux
# clean
$ make clean; ccache -C
# display distcc jobs, plz in another terminal
$ distccmon-text 2
# or, with gui
$ distccmon-gnome &
# build with distcc
$ export DISTCC_HOSTS="192.168.199.247 192.168.199.104"
$ time make -j12 CC="distcc gcc -std=gnu99" linux

Cost:

	normal	ccache	distcc	distcc + ccache
cost	4.069s	0.134s	1.301s	1.296s

P.S. distcc: 2 quad-core pc, -j12

normal	make	make -j4
cost	4.069s	1.403s

Use with ccache:

NOTE: This use of ccache is incompatible with use of distcc’s “pump” mode.

distcc on Mac

Install:

# install
$ brew install distcc
$ brew info distcc
$ distcc --help
$ distccd --help
# show ip
$ ifconfig | grep "inet " | grep -v 127.0.0.1
192.168.199.182
# config
$ vi /usr/local/Cellar/distcc/3.2rc1/homebrew.mxcl.distcc.plist
    ...
    ProgramArguments
    
        /usr/local/opt/distcc/bin/distccd
        --daemon
        --no-detach
        --allow=192.168.0.0/16
    
    ...
# start server
$ brew services start distcc
# reboot & verify
$ ps aux | grep distccd

Usage:

# download
$ curl -R -O http://www.lua.org/ftp/lua-5.3.4.tar.gz
$ tar zxf lua-5.3.4.tar.gz
$ cd lua-5.3.4
# build
$ time make macosx
# clean
$ make clean; ccache -C
# display distcc jobs, plz in another terminal
$ distccmon-text 2
# build with distcc
$ export DISTCC_HOSTS="192.168.199.182 192.168.199.105"
$ vi src/Makefile
    macosx:
            $(MAKE) $(ALL) SYSCFLAGS="-DLUA_USE_MACOSX" SYSLIBS="-lreadline" CC="distcc cc"
$ time make -j12 macosx

Note: It is recommended that using distcc splits different platforms.

dmucs on Ubuntu

DMUCS is a system that allows a group of users to share a compilation farm. Each compilation request from each user will be sent to the fastest available machine, every time.

Install:

# install
$ sudo apt-get install dmucs
# bug fix for restart on boot
#   dmucs: race condition in start scripts prevents loadavg starting
#   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=798830
$ sudo vi /etc/init.d/dmucs
    start)
          if [ "$SERVER" = "yes" ]; then
              ...
              if start_server ;  then
              ...
          if [ "$USE_SERVER" ]; then
              ...
              if start_loadavg ;  then
              ...

P.S. Unfortunately, this fix doesn’t work on my Ubuntu, although start_loadavg result is ok ==.

You could log the output of “/etc/init.d/dmucs” script, add this at the top:
1
2
exec 1>/tmp/init.log 2>&1
set -x

Server machine (e.g. 192.168.199.210):

# config
$ sudo vi /etc/default/dmucs
    SERVER=yes
    USE_SERVER=
# hosts-info
$ sudo vi /etc/dmucs.conf
    # Format: machine number-of-cpus power-index
    192.168.199.104   4   8
    192.168.199.247   4  10
# start
$ sudo /etc/init.d/dmucs start
# verify restart on boot
$ sudo reboot
$ ps aux | grep dmucs
# display which hosts/cpus are available
$ monitor

Each host (e.g. 192.168.199.104 192.168.199.247):

# config
$ sudo vi /etc/default/dmucs
    SERVER=no
    USE_SERVER=192.168.199.210
# start
$ sudo /etc/init.d/dmucs start
# verify restart on boot
$ sudo reboot
$ ps aux | grep loadavg

Usage:

$ cd lua-5.3.4
# build
$ time make -j12 CC="gethost -s 192.168.199.210 distcc gcc -std=gnu99" linux

Issue: `loadavg` get wrong host on Ubuntu 14.04, then `dmucs` not recognize the hosts

Run loadavg -s 192.168.199.210 -D, see “Writing –>127.0.1.1 …<– to the server”.

Easily avoid this issue as follows:

# comment host "127.0.1.1"
$ sudo vi /etc/hosts
    127.0.0.1   localhost
    #127.0.1.1  ubuntu
    ...
# restart loadavg
$ sudo /etc/init.d/dmucs restart

However, it will affect dnsmasq at least.

Learn more, plz google “127.0.1.1”, “ubuntu dnsmasq 127.0.1.1”.

Install Caffe & TensorFlow with Python and CUDA (macOS)

2017-04-12T09:22:25.000Z

Requirements

Python Installation

pyenv

$ brew install pyenv
$ pyenv install --list
$ pyenv install anaconda3-4.3.1
$ pyenv global anaconda3-4.3.1
$ pyenv versions
  system
* anaconda3-4.3.1 (set by /Users/John/.pyenv/version)

Enable pyenv in your shell:

1
2
3

$ vi ~/.bash_profile
# eval "$(pyenv init -)"
$ exit

$ python --version
Python 3.6.0 :: Anaconda 4.3.1 (x86_64)
$ pip --version
pip 9.0.1 from /Users/John/.pyenv/versions/anaconda3-4.3.1/lib/python3.6/site-packages (python 3.6)
$ conda --version
conda 4.3.14

CUDA Installation

For Mac OS X, please see Setup GPU for Mac

CUDA Toolkit 8.0

cuDNN v5

$ sudo mv include/cudnn.h /Developer/NVIDIA/CUDA-8.0/include/
$ sudo mv lib/libcudnn* /Developer/NVIDIA/CUDA-8.0/lib
$ sudo ln -s /Developer/NVIDIA/CUDA-8.0/lib/libcudnn* /usr/local/cuda/lib/
$ sudo ln -s /usr/local/cuda/lib/libcuda.dylib /usr/local/cuda/lib/libcuda.1.dylib

OpenCV Installation

1	$ curl -O https://github.com/opencv/opencv/archive/3.2.0.zip

$ cd opencv
$ mkdir build
$ cd build/
$ PY3_DIR=$HOME/.pyenv/versions/anaconda3-4.3.1 && \
export CPLUS_INCLUDE_PATH=$PY3_DIR/include/python3.6m && \
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D PYTHON3_LIBRARY=$PY3_DIR/lib \
-D PYTHON3_INCLUDE_DIR=$PY3_DIR/include/python3.6m \
-D PYTHON3_EXECUTABLE=$HOME/.pyenv/shims/python \
-D BUILD_opencv_python2=OFF \
-D BUILD_opencv_python3=ON \
-D BUILD_EXAMPLES=ON \
-D INSTALL_PYTHON_EXAMPLES=OFF \
-D INSTALL_C_EXAMPLES=OFF \
..
$ make -j8
$ make install

Issue fixed by export CPLUS_INCLUDE_PATH=$PY3_DIR/include/python3.6m:

[ 73%] Building CXX object modules/python3/CMakeFiles/opencv_python3.dir/__/src2/cv2.cpp.o
/Users/John/Workspace/Fever/Vision/opencv/modules/python/src2/cv2.cpp:6:10: fatal error: 'Python.h' file not
      found
#include 
         ^
1 error generated.

Caffe

Installation

1	$ brew install -vd snappy leveldb gflags glog szip lmdb

# with Python pycaffe needs dependencies built from source
$ brew install --build-from-source -vd protobuf --with-python
$ brew install --build-from-source -vd boost
$ brew install --build-from-source -vd boost-python --with-python3 --without-python

Compilation

1
2
3

$ git clone https://github.com/BVLC/caffe.git
$ cd caffe/
$ git checkout rc5

1 2	$ cp Makefile.config.example Makefile.config # Adjust Makefile.config (for example, if using Anaconda Python, or if cuDNN is desired)

USE_CUDNN := 1

OPENCV_VERSION := 3

# brew install openblas
BLAS := open
BLAS_INCLUDE := $(shell brew --prefix openblas)/include
BLAS_LIB := $(shell brew --prefix openblas)/lib

# PYTHON_INCLUDE := /usr/include/python2.7 \
        # /usr/lib/python2.7/dist-packages/numpy/core/include

ANACONDA_HOME := $(HOME)/.pyenv/versions/anaconda3-4.3.1
PYTHON_INCLUDE := $(ANACONDA_HOME)/include \
        $(ANACONDA_HOME)/include/python3.6m \
        $(ANACONDA_HOME)/lib/python3.6/site-packages/numpy/core/include

# ll /usr/local/lib/*boost*py*
PYTHON_LIBRARIES := boost_python3 python3.6m

# PYTHON_LIB := /usr/lib
PYTHON_LIB := $(ANACONDA_HOME)/lib

WITH_PYTHON_LAYER := 1

1 2	$ make all -j8 $ make pycaffe -j8

$ vi ~/.bash_profile
# export PYTHONPATH=$HOME/Workspace/Fever/DeepLearning/caffe/python:$PYTHONPATH
$ python
>>> import caffe
>>> caffe.__version__
'1.0.0-rc5'
# Or,
$ python -c "import caffe; print(caffe.__version__)"
1.0.0-rc5

Issues

Issue, `make pycaffe`:

Undefined symbols for architecture x86_64:
  "boost::python::detail::init_module(PyModuleDef&, void (*)())", referenced from:
      _PyInit__caffe in _caffe-57cf03.o
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make: *** [python/caffe/_caffe.so] Error 1

Solution:

1 2	$ brew info boost-python $ brew reinstall boost-python --with-python3 --without-python

Issue, `import caffe`:

>>> import caffe
Failed to include caffe_pb2, things might go wrong!
Traceback (most recent call last):
  File "", line 1, in 
  File "/Users/John/Workspace/Fever/DeepLearning/caffe/python/caffe/__init__.py", line 4, in 
    from .proto.caffe_pb2 import TRAIN, TEST
  File "/Users/John/Workspace/Fever/DeepLearning/caffe/python/caffe/proto/caffe_pb2.py", line 6, in 
    from google.protobuf.internal import enum_type_wrapper
ModuleNotFoundError: No module named 'google'

Solution:

1	$ pip install protobuf

# https://anaconda.org/anaconda/protobuf/files
$ conda install protobuf
Fetching package metadata .........
Solving package specifications: .

UnsatisfiableError: The following specifications were found to be in conflict:
  - protobuf -> python 2.7* -> openssl 1.0.1*
  - python 3.6*
Use "conda info " to see the dependencies for each package.

Others

How display dependencies:

1
2
3

$ cd build/lib/
# ldd -r libcaffe.so.1.0.0-rc5
$ otool -L libcaffe.so.1.0.0-rc5

TensorFlow

Installation

$ pip install tensorflow-gpu
# Or,
# tensorflow-gpu: https://pypi.python.org/pypi/tensorflow-gpu
$ export TF_BINARY_URL=https://pypi.python.org/packages/d6/54/f7cf39483b16c25a8c132dc0b81aea2ae91e367da749f17dddeaf739123a/tensorflow_gpu-1.1.0rc1-cp36-cp36m-macosx_10_11_x86_64.whl
$ pip install --upgrade $TF_BINARY_URL

$ python
...
>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
>>> print(sess.run(hello))
Hello, TensorFlow!
>>> a = tf.constant(10)
>>> b = tf.constant(32)
>>> print(sess.run(a + b))
42
>>>

Issues

Issue, `import tensorflow`:

>>> import tensorflow as tf
I tensorflow/stream_executor/dso_loader.cc:135] successfully opened CUDA library libcublas.8.0.dylib locally
I tensorflow/stream_executor/dso_loader.cc:135] successfully opened CUDA library libcudnn.5.dylib locally
I tensorflow/stream_executor/dso_loader.cc:135] successfully opened CUDA library libcufft.8.0.dylib locally
Segmentation fault: 11

Solution:

1	$ sudo ln -s /usr/local/cuda/lib/libcuda.dylib /usr/local/cuda/lib/libcuda.1.dylib

Installing Tensorflow with GPU support on os x 10.11

Issue, “Library not loaded: @rpath/libcublas.8.0.dylib”:

ImportError: dlopen(/Users/John/.pyenv/versions/anaconda3-4.3.1/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so, 10): Library not loaded: @rpath/libcublas.8.0.dylib

Solution: Disbale SIP

Reboot into “Recovery OS”,

1	$ csrutil disable

Restart into macOS,

1 2	$ python -c "import tensorflow; print(tensorflow.__version__)" 1.1.0-rc1

References

Awesome TensorFlow

使用 Djinni 开发 Android, iOS 共享库

2016-05-06T07:30:41.000Z

Djinni 是一个用来生成跨语言的类型声明和接口绑定的工具，主要用于 C++ 和 Java 以及 Objective-C 间的互通。

此文，将介绍如何使用 Djinni 开发 Android, iOS 的共享库。这会带来几个好处：

用了接口描述文件。声明清晰、修改简易，并保证了跨平台接口的一致性。
自动生成接口绑定代码。免去了绑定 C++ 和 Java (JNI) 及 Objective-C (Objective-C++) 的麻烦。

初见：Djinni 及其样例

下载 Djinni

1	$ git clone https://github.com/dropbox/djinni.git

编译 Djinni

1 2	$ cd [djinni_root]/ $ src/build

于[djinni_root]/src/support/sbt.resolvers.properties内可添加镜像源。

sbt下载慢？

使用 Djinni

生成样例接口代码：

1 2	$ cd [djinni_root]/example/ # example_root $ ./run_djinni.sh

即会生成到djinni-output-temp临时目录，最终复制到generated-src生成目录。

这里可以看到：依据描述文件example.djinni，C++ 和 Java 及 Objective-C 的绑定代码都会自动生成好。继续要做的，只是写它们的具体实现，见样例的handwritten-src目录。

如果要清除输出目录：

1	$ ./run_djinni.sh clean

编译样例

[djinni_root]/Makefile已配置好了依赖，执行相应目标即可。

1	$ cd [djinni_root]/

注：下载好 Djinni ，即可开始编译样例了。

编译 Android 工程：

1	$ make example_android

样例的 Android 工程在[example_root]/android目录，动态库生成在[example_root]/android/app/libs目录。或者，利用 Android Studio / Gradle 来运行编译。

编译 iOS 工程：

1	$ make example_ios

样例的 iOS 工程在[example_root]/objc目录， lib 工程生成在[djinni_root]\build_ios目录。然后，可以打开[example_root]/objc/TextSort.xcworkspace来运行编译。

如果要清理工程：

1	$ make clean

准备 GYP

编译样例时， Android NDK 与 iOS 的 Library 工程都需依赖 GYP 生成。 make 时，会自行 clone 到[djinni_root]/deps/gyp目录。

1	$ git clone https://chromium.googlesource.com/external/gyp.git

GYP 生成 Android Makefile 时，目前会遇到如下错误：

ImportError: No module named android

所以，需要切换到旧版本。此后的那个 commit 移除了 Android 的生成器。

1 2	$ cd gyp/ $ git checkout -q 0bb67471bca068996e15b56738fa4824dfa19de0

注：Stop using gyp’s android generator

从无到有：Hello Djinni

C++ 接口

定义接口描述文件

hellodjinni.djinnihellodjinni.djinni

hello_djinni = interface +c {
    static create(): hello_djinni;
    get_hello_djinni(): string;
}

生成接口绑定代码

写了个简单的 Shell 脚本来执行 Djinni 命令，如下：

run_djinni.sh:run_djinni.sh

#!/bin/bash
set -e
shopt -s nullglob
base_dir=$(cd "`dirname "$0"`" && pwd)
# Read local properties
eval $(cat $base_dir/local.properties | sed 's/<[^>]*>//g' | sed 's/\./_/g')
if [[ -z $djinni_dir ]]; then
    echo "Unspecified djinni.dir in local.properties" 1>&2
    exit 1
fi
out_dir="$base_dir/generated-src"
cpp_out="$out_dir/cpp"
jni_out="$out_dir/jni"
objc_out="$out_dir/objc"
java_out="$out_dir/java/cc/eevee/hellodjinni"
java_package="cc.eevee.hellodjinni"
cpp_namespace="hellodjinni"
objc_type_prefix="HD"
djinni_file="$base_dir/hellodjinni.djinni"
[[ -e $out_dir ]] && rm -rf $out_dir
$djinni_dir/src/run-assume-built \
    --java-out $java_out \
    --java-package $java_package \
    --ident-java-field mFooBar \
    \
    --cpp-out $cpp_out \
    --cpp-namespace $cpp_namespace \
    \
    --jni-out $jni_out \
    --ident-jni-class NativeFooBar \
    --ident-jni-file NativeFooBar \
    \
    --objc-out $objc_out \
    --objcpp-out $objc_out \
    --objc-type-prefix $objc_type_prefix \
    \
    --idl $djinni_file

其读取了local.properties内配置的 Djinni 目录路径。

local.propertieslocal.properties

1
2
3

djinni.dir=
gyp.dir=
ndk.dir=

运行后，代码生成在了generated-src目录。

1	$ ./run_djinni.sh

实现 C++ 接口

首先，创建src目录，存放手写的代码。然后，于子目录cpp内实现 C++ 接口。

src/cpp/hello_djinni_impl.hpphello_djinni_impl.hpp

#pragma once
#include "hello_djinni.hpp"
namespace hellodjinni {
class HelloDjinniImpl : public HelloDjinni {
public:
    // Constructor
    HelloDjinniImpl();
    // Our method that returns a string
    std::string get_hello_djinni();
};
}  // namespace hellodjinni

src/cpp/hello_djinni_impl.cpphello_djinni_impl.cpp

#include "hello_djinni_impl.hpp"
#include 
using namespace hellodjinni;
std::shared_ptr HelloDjinni::create() {
    return std::make_shared();
}
HelloDjinniImpl::HelloDjinniImpl() {
}
std::string HelloDjinniImpl::get_hello_djinni() {
    std::string result = "Hello Djinni! ";
    time_t t = time(0);
    tm now = *localtime(&t);
    char tm_desc[200] = {0};
    if (strftime(tm_desc, sizeof(tm_desc)-1, "%r", &now)>0) {
        result += tm_desc;
    }
    return result;
}

C++ 工程

这里用 XCode 创建一个 C++ 工程，来测试 C++ 接口代码。

首先，打开 XCode ，选择”Create a new Xcode project”。然后，选择”Command Line Tool”，来新建命令行工具。

“Next”到下一步时，”Language”选择”C++”。

工程最后保存到了project/cpp目录。整个文件结构如下：

接下来，把以下 C++ 接口代码文件，拖动到 XCode 工程目录来引入。

1
2
3

generated-src/cpp/hello_djinni.hpp
src/cpp/hello_djinni_impl.cpp
src/cpp/hello_djinni_impl.hpp

注：取消”Copy items if needed”，选择”Create folder references”。只需引用文件，避免复制。

然后，编写好main.cpp的代码：

project/cpp/HelloDjinni/HelloDjinni/main.cppmain.cpp

#include 
#include "hello_djinni_impl.hpp"
int main(int argc, const char * argv[]) {
    typedef hellodjinni::HelloDjinni HelloDjinni;
    auto hd = HelloDjinni::create();
    auto result = hd->get_hello_djinni();
    std::cout << result << std::endl;
    return 0;
}

最后，运行项目，结果如下：

或者，写个project/Cpp.mk：

project/Cpp.mkCpp.mk

MD := -mkdir -p
RD := -rm -rf
RM := -rm -f
CC := gcc
CXX := g++
CXXFLAGS := -std=c++11 -Wall
ifdef DEBUG
CXXFLAGS += -g -DDEBUG
else
CXXFLAGS += -O2 -DNDEBUG
endif
OUT_DIR ?= build
CPP_OUT ?= $(OUT_DIR)/cpp
CPP_INCLUDES := \
    ../generated-src/cpp \
    ../src/cpp
CPP_SOURCES := \
    ../src/cpp/hello_djinni_impl.cpp \
    cpp/HelloDjinni/HelloDjinni/main.cpp
CPP_TARGET := $(CPP_OUT)/HelloDjinni
all: cpp_pro
clean:
    $(RD) $(CPP_OUT)/
cpp_pro: $(CPP_SOURCES)
    @echo "\033[1;35;47mBuild cpp project...\033[0m"
    @$(MD) $(CPP_OUT)
    $(CXX) $(CXXFLAGS) $(CPP_SOURCES) -o $(CPP_TARGET) \
        $(foreach d, $(CPP_INCLUDES), -I$d)
    @echo "\033[32mOutput:\033[0m\n$(CPP_TARGET)"
    @echo "\033[32mRunning:\033[0m"
    @$(CPP_TARGET)
.PHONY: cpp_pro clean all

然后，make -f Cpp.mk编译运行，结果如下：

Build cpp project...
g++ -std=c++11 -Wall -O2 -DNDEBUG \
    ../src/cpp/hello_djinni_impl.cpp \
    cpp/HelloDjinni/HelloDjinni/main.cpp \
    -o build/cpp/HelloDjinni \
    -I../generated-src/cpp -I../src/cpp
Output:
build/cpp/HelloDjinni
Running:
Hello Djinni! 11:46:56 PM

iOS 工程

打开 XCode，”File > New > Workspace”新建一个工作区，保存到project/ios目录。

接着，”File > New > Project”，选择”Single View Application”，创建 iOS 工程。

“Next”到下一步时，”Language”选择”Objective-C”。

工程保存到project/ios目录，”Add to”选择刚才的工作区。

“Create”完成创建。

生成接口 Libraries 工程

利用 Djinni, GYP 及 Make 生成接口 Libraries 工程。

首先，创建 GYP 文件：

project/libhellodjinni.gyplibhellodjinni.gyp

{
  "targets": [
    {
      "target_name": "libhellodjinni_jni",
      "type": "shared_library",
      "dependencies": [
        "<(DJINNI_DIR)/support-lib/support_lib.gyp:djinni_jni",
      ],
      "ldflags": [ "-llog", "-Wl,--build-id,--gc-sections,--exclude-libs,ALL" ],
      "sources": [
        "<(DJINNI_DIR)/support-lib/jni/djinni_main.cpp",
        ",
        ",
        ",
      ],
      "include_dirs": [
        "../generated-src/jni",
        "../generated-src/cpp",
        "../src/cpp",
      ],
    },
    {
      "target_name": "libhellodjinni_objc",
      "type": "static_library",
      "dependencies": [
        "<(DJINNI_DIR)/support-lib/support_lib.gyp:djinni_objc",
      ],
      "sources": [
        ",
        ",
        ",
      ],
      "include_dirs": [
        "../generated-src/objc",
        "../generated-src/cpp",
        "../src/cpp",
      ],
    },
  ],
}

注意：
1) sources内的路径必须是相对路径。虽然会识别以/开头的字符串为绝对路径，但在 XCode 工程内其路径引用是不正确的。
2) GYP 生成 Android Makefile 时，运行命令时的工作目录，必须能够直接子目录到所有代码，包括依赖的 Djinni 的 support-lib 。不然，会报如下错误：

AssertionError: Path %s attempts to escape from gyp path %s !)

👌的话，GypAndroid.mk会生成到当前工作目录。GYP 生成 Android 时，不允许指定--generator-output：

AssertionError: The Android backend does not support options.generator_output.

所以，简单的解决办法是，文件结构与[djinni_root]/example一致，并git submodule Djinni 与 GYP 到工程目录内。

1
2
3

$ git init
$ git submodule add https://github.com/dropbox/djinni.git deps/djinni
$ git submodule add https://chromium.googlesource.com/external/gyp.git deps/gyp

如果仍想要 Djinni 与 GYP 独立于工程目录外，同时又能够工作在工程目录，那么需要把依赖的东西复制进工程目录。之后，即是这样做的。

接下来，创建 Makefile 文件：

project/MakefileMakefile

MD := -mkdir -p
RD := -rm -rf
RM := -rm -f
SED_PROP := "s/<[^>]*>//g;s/.*=\(.*\)/\1/"
define read_prop
    $(eval $(1) := $(shell ../tools/read_properties.sh \
        | grep -m1 $(2) | sed $(SED_PROP)))
endef
$(call read_prop,GYP_DIR,gyp_dir)
$(call read_prop,NDK_DIR,ndk_dir)
OUT_DIR ?= build
IOS_OUT ?= $(OUT_DIR)/ios
DEPS_DIR ?= deps
DEPS_DJINNI_DIR := $(DEPS_DIR)/djinni
DJINNI_OUT_DIR ?= ../generated-src
.PHONY: all
all: cpp_pro ios_pro android_pro android_pro2
.PHONY: clean
clean:
    @make ios_pro_clean
    @make android_pro_clean
    @make android_pro2_clean
    @echo "\033[1;35;47mClean others...\033[0m"
    $(RD) $(OUT_DIR)/
    $(RD) $(DEPS_DIR)/
    $(RD) $(DJINNI_OUT_DIR)/
    $(RM) ../GypAndroid.mk
    $(RM) libhellodjinni_jni.target.mk
    @# @make cpp_pro_clean
deps:
    @echo "\033[1;35;47mPrepare dependencies...\033[0m"
    @../tools/prepare_deps.sh $(DEPS_DIR)
$(DJINNI_OUT_DIR): ../hellodjinni.djinni
    @echo "\033[1;35;47mGenerate djinni interface code...\033[0m"
    @../run_djinni.sh
.PHONY: djinni
djinni: $(DJINNI_OUT_DIR)
.PHONY: cpp_pro
cpp_pro: $(DJINNI_OUT_DIR)
    @make -f Cpp.mk cpp_pro
.PHONY: cpp_pro_clean
cpp_pro_clean:
    @make -f Cpp.mk clean
$(IOS_OUT)/libhellodjinni.xcodeproj: deps $(DJINNI_OUT_DIR) libhellodjinni.gyp \
        $(DEPS_DJINNI_DIR)/support-lib/support_lib.gyp
    @echo "\033[1;35;47mGenerate libhellodjinni.xcodeproj...\033[0m"
    @$(GYP_DIR)/gyp --depth=. -f xcode -DOS=ios \
        --generator-output $(IOS_OUT) \
        -DDJINNI_DIR=$(DEPS_DJINNI_DIR) \
        -I$(DEPS_DJINNI_DIR)/common.gypi \
        libhellodjinni.gyp
.PHONY: ios_pro
ios_pro: $(IOS_OUT)/libhellodjinni.xcodeproj
    @echo "\033[1;35;47mBuild ios project...\033[0m"
    xcodebuild -workspace ios/HelloDjinni.xcworkspace \
        -scheme HelloDjinni -configuration 'Debug' -sdk iphonesimulator
.PHONY: ios_pro_clean
ios_pro_clean:
    @echo "\033[1;35;47mClean ios project...\033[0m"
    @-xcodebuild -workspace ios/HelloDjinni.xcworkspace \
        -scheme HelloDjinni -configuration 'Debug' -sdk iphonesimulator clean
../GypAndroid.mk: deps $(DJINNI_OUT_DIR) libhellodjinni.gyp \
        $(DEPS_DJINNI_DIR)/support-lib/support_lib.gyp
    @echo "\033[1;35;47mGenerate GypAndroid.mk...\033[0m"
    @make gyp_android
    @cd .. && ANDROID_BUILD_TOP=$(NDK_DIR) \
        $(GYP_DIR)/gyp --depth=. -f android -DOS=android \
        -DDJINNI_DIR=$(DEPS_DJINNI_DIR) \
        -Iproject/$(DEPS_DJINNI_DIR)/common.gypi \
        project/libhellodjinni.gyp \
        --root-target=libhellodjinni_jni
    @make gyp_master
.PHONY: android_pro
android_pro: ../GypAndroid.mk
    @echo "\033[1;35;47mBuild android project (HelloDjinni)...\033[0m"
    cd android/HelloDjinni/ && ./gradlew app:assembleDebug
    @echo "\033[32mApks produced at:\033[0m"
    @python $(DEPS_DJINNI_DIR)/example/glob.py \
        android/HelloDjinni/app/build/outputs/apk/ '*.apk'
.PHONY: android_pro_clean
android_pro_clean: ../GypAndroid.mk
    @echo "\033[1;35;47mClean android project (HelloDjinni)...\033[0m"
    @ndk-build -C android/HelloDjinni/app-core clean
    @-cd android/HelloDjinni/ && ./gradlew clean
.PHONY: android_pro2
android_pro2: deps djinni libhellodjinni.gyp \
        $(DEPS_DJINNI_DIR)/support-lib/support_lib.gyp
    @echo "\033[1;35;47mBuild android project (HelloDjinni2)...\033[0m"
    cd android/HelloDjinni2/ && ./gradlew app:assembleDebug
    @echo "\033[32mApks produced at:\033[0m"
    @python $(DEPS_DJINNI_DIR)/example/glob.py \
        android/HelloDjinni2/app/build/outputs/apk/ '*.apk'
.PHONY: android_pro2_clean
android_pro2_clean:
    @echo "\033[1;35;47mClean android project (HelloDjinni2)...\033[0m"
    @-cd android/HelloDjinni2/ && ./gradlew clean
.PHONY: gyp_status
gyp_status:
    @cd $(GYP_DIR) && git status
.PHONY: gyp_master
gyp_master:
    @echo "\033[1;35;47mCheckout gyp to master...\033[0m"
    @cd $(GYP_DIR) && git checkout master
.PHONY: gyp_android
gyp_android:
    @echo "\033[1;35;47mCheckout gyp to android generator...\033[0m"
    @cd $(GYP_DIR) && git checkout -q 0bb67471bca068996e15b56738fa4824dfa19de0
.PHONY: test_vars
test_vars:
    @echo "\033[1;35;47mPrint variables...\033[0m"
    @echo GYP_DIR=$(GYP_DIR)
    @echo NDK_DIR=$(NDK_DIR)
    @echo OUT_DIR=$(OUT_DIR)
    @echo IOS_OUT=$(IOS_OUT)
    @echo DEPS_DIR=$(DEPS_DIR)
    @echo DEPS_DJINNI_DIR=$(DEPS_DJINNI_DIR)

其也包括了 Android 工程的配置。

注：额外依赖了两个辅助脚本，说明如下：
read_properties.sh，读取local.properties内配置的路径。
prepare_deps.sh，准备依赖的文件到指定目录。

然后，运行生成 XCode 的 libhellodjinni 工程。

1 2	$ cd project/ $ make ios_pro

其生成在project/build/ios目录。 XCode 直接打开libhellodjinni.xcodeproj，即可选择目标进行编译。

但可能此时 libhellodjinni_jni 与 djinni_jni (support_lib) 目标，不能够找到jni.h。由于 ios 上不需要 jni 绑定，后续也不需依赖，没多大影响。

也让其可通过编译的话，只需要选择目标，在”Build Settings > Header Search Paths”内，添加 Java VM include 就好。

运行如下命令，获得当前的 Java 头文件路径：

1	$ ls -l `which java` \| sed 's/^.-> $.Current$.$/\1\/Headers/;q'

结果，如：”/System/Library/Frameworks/JavaVM.framework/Versions/Current/Headers”。

或者，这里找：”/Library/Java/JavaVirtualMachines/jdk1.8.0_25.jdk/Contents/Home/include/“。

添加接口 Libraries 依赖

现在，给 iOS 工程上添加上接口 Libraries 的依赖。

首先，打开先前的工作区project/ios/HelloDjinni.xcworkspace。于左侧项目导航的灰色区域，”Ctrl+Click”或右击打开菜单。选择”Add Files to “HelloDjinni””，添加生成好的libhellodjinni.xcodeproj和support_lib.xcodeproj两个库工程。

之后，项目导航选中”HelloDjinni”工程，并选择”HelloDjinni”目标。在”Build Phases”标签页的”Link Binaries With Libraries”选项下，新增 libhellodjinni_objc.a 与 libdjinni_objc.a 的依赖，如下：

在”Build Settings”标签页，找到”Header Search Paths”，添加头文件搜索路径：

1 2	$(SRCROOT)/../../deps/djinni/support-lib/objc $(SRCROOT)/../../../generated-src/objc

为了兼容 Objective-C++ 桥接代码，需要将HelloWorld/Supporting Files/main.m重命名为main.mm。

最终，工作区会类似于下面这样：

完成 UI 并运行

于ViewController.m内编写代码，创建 UI 并调用接口代码。

project/ios/HelloDjinni/HelloDjinni/ViewController.mViewController.m

#import "ViewController.h"
#import "HDHelloDjinni.h"
@interface ViewController ()
@end
@implementation ViewController {
    HDHelloDjinni *_helloDjinniInterface;
    UIButton *_button;
    UITextView *_textView;
}
- (void)viewDidLoad {
    [super viewDidLoad];
    // instantiate our library interface
    _helloDjinniInterface = [HDHelloDjinni create];
    // create a button programatically for the demo
    _button = [UIButton buttonWithType:UIButtonTypeRoundedRect];
    [_button addTarget:self
                action:@selector(buttonWasPressed:)
      forControlEvents:UIControlEventTouchUpInside];
    [_button setTitle:@"Get Hello Djinni!" forState:UIControlStateNormal];
    _button.frame = CGRectMake(20.0, 20.0, 280.0, 40.0);
    [self.view addSubview:_button];
    // create a text view programatically
    _textView = [[UITextView alloc] init];
    // x, y, width, height
    _textView.frame = CGRectMake(20.0, 80.0, 280.0, 380.0);
    [self.view addSubview:_textView];
}
- (void)buttonWasPressed:(UIButton *)sender {
    NSString *response = [_helloDjinniInterface getHelloDjinni];
    _textView.text = [NSString stringWithFormat:@"%@\n%@", response, _textView.text];
}
- (void)didReceiveMemoryWarning {
    [super didReceiveMemoryWarning];
    // Dispose of any resources that can be recreated.
}
@end

“Product > Run”运行项目。 UI 上”Get Hello Djinni!”的按钮，每点击下就会添加条从 C++ 返回的信息。

Android 工程

Android 工程介绍了两种方式，来整合 NDK Library ：

一是，使用 GYP 生成的 Android Makefile ， Gradle 配置 ndk-build 进行编译。
二是，使用 Experimental Plugin ，直接配置成支持 NDK 的工程。

使用 GypMakefile

GYP 生成 Android Makefile ，使用之前写好的project/Makefile：

1 2	$ cd project/ $ make android_pro

GypAndroid.mk会生成到父级hellodjinni目录。

如果还没准备好 Studio 工程，不会继续生成 APK ，会报“找不到android/HelloDjinni/”。

现在，打开 Android Studio，选择”Start a new Android Studio Project”。

“New”页，”Project Location”存到project/android/HelloDjinni，如下：

之后，”Target”页选”Phone and Tablet”，”Add”页选”Empty Activity”，最终完成新建。

于”File > Project Structure > SDK Location > Android NDK Location”，设置 NDK 路径：

接下来，独立建一个 app-core Library 模块来引用 C++ 代码。”File > New Module”，选”Android Library”：

然后，修改此 app-core 的build.gradle，添加引用及 NDK 编译。变更如下：

project/android/HelloDjinni/app-core/build.gradlebuild.gradle

apply plugin: 'com.android.library'
android {
    ...
    sourceSets {
        main {
            java.srcDirs += ['../../../../generated-src/java']
            jni.srcDirs = []
            jniLibs.srcDirs = ['libs']
        }
    }
}
dependencies {
    ...
    //compile 'com.android.support:appcompat-v7:23.3.0'
}
task ndkBuild(type: Exec) {
    def ndkDir = plugins.getPlugin('com.android.library').sdkHandler.ndkFolder
    if (ndkDir == null) {
        def gradle_project_root = project.rootProject.rootDir
        throw new GradleException("NDK is not configured. Make sure there is a local.properties " +
                "file with an ndk.dir entry in the directory ${gradle_project_root}.")
    }
    def ndkBuildExecutable = new File(ndkDir, 'ndk-build')
    if (!ndkBuildExecutable.exists()) {
        throw new GradleException("Could not find ndk-build. The configured NDK directory ${ndkDir} may not be correct.")
    }
    environment("NDK_PROJECT_PATH", projectDir)
    environment("GYP_CONFIGURATION", "Release")
    commandLine ndkBuildExecutable
}
tasks.withType(JavaCompile) {
    compileTask -> compileTask.dependsOn ndkBuild
}

项目导航栏切到 Project 视图，在 app-core 下新建jni目录，创建 NDK 的工程文件。

project/android/HelloDjinni/app-core/jni/Android.mkAndroid.mk

1
2
3

# always force this build to re-run its dependencies
FORCE_GYP := $(shell make -C ../../../../ GypAndroid.mk)
include ../../../../GypAndroid.mk

project/android/HelloDjinni/app-core/jni/Application.mkApplication.mk

# Android makefile for libhellodjinni shared lib
# Application.mk: http://developer.android.com/ndk/guides/application_mk.html
# APP_ABI := all
# skipping mips / mips64
APP_ABI := armeabi armeabi-v7a arm64-v8a x86 x86_64
APP_OPTIM := release
APP_PLATFORM := android-14
# GCC 4.9 Toolchain - requires NDK r10
NDK_TOOLCHAIN_VERSION = 4.9
# GNU libc++ is the only Android STL which supports C++11 features
APP_CFLAGS += -Wall
APP_CPPFLAGS += -std=c++11 -frtti -fexceptions
APP_STL := gnustl_static
APP_BUILD_SCRIPT := jni/Android.mk
APP_MODULES := libhellodjinni_jni

这样，独立的 app-core Library 就👌了。

回到 app ，修改其build.gradle，以依赖 app-core 。变更如下：

project/android/HelloDjinni/app/build.gradlebuild.gradle

dependencies {
    ...
    compile project(':app-core')
}

新建MyApplication.java，作为自定义 Application。并设置到AndroidManifest.xml内”application”的”name”字段。

project/android/HelloDjinni/app/src/.../MyApplication.javaMyApplication.java

package cc.eevee.hellodjinni;
import android.app.Application;
public class MyApplication extends Application {
    static {
        try {
            System.loadLibrary("hellodjinni_jni");
        } catch (UnsatisfiedLinkError e) {
            System.err.println("Native code library failed to load.\n" + e);
        }
    }
    @Override
    public void onCreate() {
        super.onCreate();
    }
}

最终，项目导航 Android 和 Project 视图类似下面这样：

接着，修改 app UI，MainActivity.java及其布局activity_main.xml：

project/android/HelloDjinni/app/src/.../MainActivity.javaMainActivity.java

package cc.eevee.hellodjinni;
import android.support.v7.app.AppCompatActivity;
import android.os.Bundle;
import android.view.View;
import android.widget.ScrollView;
import android.widget.TextView;
public class MainActivity extends AppCompatActivity {
    private TextView mTextView;
    private ScrollView mScrollView;
    private HelloDjinni mHelloDjinniInterface = HelloDjinni.create();
    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
        mTextView = (TextView) findViewById(R.id.textView);
        mScrollView = (ScrollView) findViewById(R.id.scrollView);
    }
    public void onButtonClick(View view) {
        mTextView.append(mHelloDjinniInterface.getHelloDjinni() + "\n");
        mScrollView.fullScroll(ScrollView.FOCUS_DOWN);
    }
}

project/android/HelloDjinni/app/src/.../activity_main.xmlactivity_main.xml


<LinearLayout
    xmlns:android="http://schemas.android.com/apk/res/android"
    xmlns:tools="http://schemas.android.com/tools"
    android:layout_width="match_parent"
    android:layout_height="match_parent"
    android:orientation="vertical"
    android:paddingBottom="@dimen/activity_vertical_margin"
    android:paddingLeft="@dimen/activity_horizontal_margin"
    android:paddingRight="@dimen/activity_horizontal_margin"
    android:paddingTop="@dimen/activity_vertical_margin"
    tools:context="cc.eevee.hellodjinni.MainActivity">
    <Button
        android:id="@+id/button"
        android:layout_width="match_parent"
        android:layout_height="wrap_content"
        android:text="Get Hello Djinni!"
        android:gravity="center"
        android:onClick="onButtonClick"/>
    <ScrollView
        android:id="@+id/scrollView"
        android:layout_width="match_parent"
        android:layout_height="match_parent">
        <TextView
            android:id="@+id/textView"
            android:layout_width="match_parent"
            android:layout_height="wrap_content"/>
    ScrollView>
LinearLayout>

“Run > Run ‘app’”运行项目。 UI 上”Get Hello Djinni!”的按钮，每点击下就会添加条从 C++ 返回的信息。

使用新试验性插件

Android Studio 1.3 版本开始支持 NDK，需要使用 Experimental Plugin 。这里为当前最新的 0.7.0 版本。

同样，打开 Android Studio，选择”Start a new Android Studio Project”，新建一个”HelloDjinni2”工程。

之后，”Target”页选”Phone and Tablet”，”Add”页选”Empty Activity”，最终完成新建。

于”File > Project Structure > SDK Location > Android NDK Location”，设置 NDK 路径。

接下来，修改成 Experimental Plugin 。先是工程配置：

project/android/HelloDjinni2/build.gradlebuild.gradle

// Top-level build file where you can add configuration options common to all sub-projects/modules.
buildscript {
    repositories {
        jcenter()
    }
    dependencies {
        classpath 'com.android.tools.build:gradle-experimental:0.7.0'
        // NOTE: Do not place your application dependencies here; they belong
        // in the individual module build.gradle files
    }
}
allprojects {
    repositories {
        jcenter()
    }
}
task clean(type: Delete) {
    delete rootProject.buildDir
}

接着是 app 模块配置：

project/android/HelloDjinni2/app/build.gradlebuild.gradle

apply plugin: 'com.android.model.application'
model {
    android {
        compileSdkVersion = 23
        buildToolsVersion = '23.0.3'
        defaultConfig {
            applicationId = 'cc.eevee.hellodjinni2'
            minSdkVersion.apiLevel    = 14
            targetSdkVersion.apiLevel = 23
            versionCode = 1
            versionName = '1.0'
        }
        buildTypes {
            release {
                minifyEnabled = false
                proguardFiles.add(file('proguard-rules.pro'))
            }
        }
    }
}
dependencies {
    compile fileTree(dir: 'libs', include: ['*.jar'])
    testCompile 'junit:junit:4.12'
    compile 'com.android.support:appcompat-v7:23.3.0'
    //compile project(':app-core')
}

这样， Experimental Plugin 就修改完成了。

接下来，仍旧独立建一个 app-core Library 模块来引用 C++ 代码。”File > New Module”，选”Android Library”：

然后，修改此 app-core 的build.gradle，支持 Experimental Plugin 并配置 NDK 。如下：

project/android/HelloDjinni2/app-core/build.gradlebuild.gradle

apply plugin: 'com.android.model.library'
model {
    android {
        compileSdkVersion = 23
        buildToolsVersion = '23.0.3'
        defaultConfig {
            minSdkVersion.apiLevel    = 14
            targetSdkVersion.apiLevel = 23
        }
        ndk {
            moduleName = 'hellodjinni'
            platformVersion = 14
            toolchain = 'gcc'
            toolchainVersion = '4.9'
            stl = 'gnustl_shared'
            CFlags.addAll(['-Wall', '-Werror'])
            cppFlags.addAll(['-std=c++11', '-fexceptions', '-frtti'])
            cppFlags.addAll([
                    "-I${file('../../../deps/djinni/support-lib')}".toString(),
                    "-I${file('../../../deps/djinni/support-lib/jni')}".toString(),
                    "-I${file('../../../../generated-src/cpp')}".toString(),
                    "-I${file('../../../../generated-src/jni')}".toString(),
            ])
            ldLibs.addAll(['log'])
            abiFilters.addAll(['armeabi', 'armeabi-v7a', 'arm64-v8a', 'x86', 'x86_64'])
        }
        sources {
            //noinspection GroovyAssignabilityCheck
            main {
                jni {
                    source {
                        srcDirs += [
                            '../../../deps/djinni/support-lib/jni',
                            '../../../../generated-src/cpp',
                            '../../../../generated-src/jni',
                            '../../../../src/cpp',
                        ]
                    }
                }
                java {
                    source {
                        srcDirs += [
                            '../../../../generated-src/java',
                        ]
                    }
                }
            }
        }
        buildTypes {
            release {
                minifyEnabled = false
                proguardFiles.add(file('proguard-rules.pro'))
            }
        }
    }
}
dependencies {
    compile fileTree(dir: 'libs', include: ['*.jar'])
    testCompile 'junit:junit:4.12'
    //compile 'com.android.support:appcompat-v7:23.3.0'
}

这样之后，可以在 app-core 下看得jni目录，包括了所有 C++ 代码。

回到 app ，修改其build.gradle，以依赖 app-core 。变更如下：

project/android/HelloDjinni2/app/build.gradlebuild.gradle

dependencies {
    ...
    compile project(':app-core')
}

新建MyApplication.java，作为自定义 Application。并设置到AndroidManifest.xml内”application”的”name”字段。

project/android/HelloDjinni2/app/src/.../MyApplication.javaMyApplication.java

package cc.eevee.hellodjinni2;
import android.app.Application;
public class MyApplication extends Application {
    static {
        try {
            System.loadLibrary("gnustl_shared");
            System.loadLibrary("hellodjinni");
        } catch (UnsatisfiedLinkError e) {
            System.err.println("Native code library failed to load.\n" + e);
        }
    }
    @Override
    public void onCreate() {
        super.onCreate();
    }
}

最终，项目导航 Android 和 Project 视图类似下面这样：

接着，修改 app UI，MainActivity.java及其布局activity_main.xml：

project/android/HelloDjinni2/app/src/.../MainActivity.javaMainActivity.java

package cc.eevee.hellodjinni2;
import android.support.v7.app.AppCompatActivity;
import android.os.Bundle;
import android.view.View;
import android.widget.ScrollView;
import android.widget.TextView;
import cc.eevee.hellodjinni.HelloDjinni;
public class MainActivity extends AppCompatActivity {
    private TextView mTextView;
    private ScrollView mScrollView;
    private HelloDjinni mHelloDjinniInterface = HelloDjinni.create();
    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
        mTextView = (TextView) findViewById(R.id.textView);
        mScrollView = (ScrollView) findViewById(R.id.scrollView);
    }
    public void onButtonClick(View view) {
        mTextView.append(mHelloDjinniInterface.getHelloDjinni() + "\n");
        mScrollView.fullScroll(ScrollView.FOCUS_DOWN);
    }
}

project/android/HelloDjinni2/app/src/.../activity_main.xmlactivity_main.xml


<LinearLayout
    xmlns:android="http://schemas.android.com/apk/res/android"
    xmlns:tools="http://schemas.android.com/tools"
    android:layout_width="match_parent"
    android:layout_height="match_parent"
    android:orientation="vertical"
    android:paddingBottom="@dimen/activity_vertical_margin"
    android:paddingLeft="@dimen/activity_horizontal_margin"
    android:paddingRight="@dimen/activity_horizontal_margin"
    android:paddingTop="@dimen/activity_vertical_margin"
    tools:context="cc.eevee.hellodjinni2.MainActivity">
    <Button
        android:id="@+id/button"
        android:layout_width="match_parent"
        android:layout_height="wrap_content"
        android:text="Get Hello Djinni!"
        android:gravity="center"
        android:onClick="onButtonClick"/>
    <ScrollView
        android:id="@+id/scrollView"
        android:layout_width="match_parent"
        android:layout_height="match_parent">
        <TextView
            android:id="@+id/textView"
            android:layout_width="match_parent"
            android:layout_height="wrap_content"/>
    ScrollView>
LinearLayout>

“Run > Run ‘app’”运行项目。 UI 上”Get Hello Djinni!”的按钮，每点击下就会添加条从 C++ 返回的信息。

NDK 参考

结语：开始使用 Djinni 吧

附：源码

Hello Djinni 的源码，这样得到：

1 2	$ git clone https://github.com/joinAero/XCalculator.git $ cd sample/hellodjinni/

修改local.properties设好环境。执行make编译，make clean清理。

文件结构：

hellodjinni/
├─project/
│  ├─android/
│  │  ├─HelloDjinni/                # Android Project with GYP & ndk-build
│  │  └─HelloDjinni2/               # Android Project with Experimental Plugin
│  ├─cpp/
│  │  └─HelloDjinni/                # Cpp Test Project
│  └─ios/
│      ├─HelloDjinni/
│      └─HelloDjinni.xcworkspace/   # iOS Project Workspace
├─src/
│  └─cpp/                           # Cpp Interface Impls
├─tools/                            # Helper Scripts
└─local.properties                  # Local Properties

附：运行环境

# OS
ProductName: Mac OS X ProductVersion: 10.11.4 BuildVersion: 15E65
# XCode
Xcode 7.3.1 Build version 7D1014
# XCode 命令行工具
# xcode-select --install
xcode-select version 2343.
# Java
java version "1.8.0_25"
Java(TM) SE Runtime Environment (build 1.8.0_25-b17)
Java HotSpot(TM) 64-Bit Server VM (build 25.25-b02, mixed mode)
# Android Studio
# Android Studio > About Android Studio
# Android Studio > Appearance & Behavior > System Settings > Updates
Android Studio 2.1
Build #AI-143.2790544
# Android NDK
GNU Make 3.81 Copyright (C) 2006 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. This program built for i386-apple-darwin11.3.0

如果未添加过 Android 环境变量，请于 ~/.bash_profile 文件内设置：

export ANDROID_HOME=$HOME/Develop/android-sdk
export PATH=$PATH:$ANDROID_HOME/tools:$ANDROID_HOME/platform-tools
export ANDROID_NDK_HOME=$HOME/Develop/android-ndk
export PATH=$PATH:$ANDROID_NDK_HOME

终端运行如下命令可立即生效：

1	$ source ~/.bash_profile

验证 SDK 与 NDK 命令行工具可用：

1
2
3

$ which android
$ which ndk-build

验证 XCode 命令行工具可用：

1
2
3

$ which xcrun
$ which xcodebuild

Flower Crown Eevee

Use Kinect on Firefly-RK3399

Preparation

Install libfreenect

Install OpenNI

Build OpenCV with OpenNI

Usage

C++

References

Issues

Failed to use libfreenect form sudo apt-get install freenect

Failed to build libfreenect on Firefly-RK3399 (aarch64)

Failed to use OpenNI2 from sudo apt-get install libopenni2-dev

Failed to use OpenNI2 from OpenNI-Linux-Arm-2.2.0.33.tar.bz2

Failed to use libfreenect with OpenNI2 (from source)

Build OpenCV on Firefly-RK3399

Install

Required Packages

Getting OpenCV Source Code

Building OpenCV from Source

Extra Links

Examples

Appendix

Install pip

Headless Firefly-RK3399 Setup

Setup Firefly

Screen Sharing from Ubuntu to Mac

Ubuntu

Mac

References

Headless Raspberry Pi Setup

Install

Prepare

Vim

Samba

Python

Then

Usage

Build OpenCV for Android with CUDA

Prerequisites

Prepare

Build

Usage

Issues

Issue when cmake

Issue when make

References

Use ccache, distcc, dmucs to accelerate builds

distcc on Ubuntu

distcc on Mac

dmucs on Ubuntu

Issue: loadavg get wrong host on Ubuntu 14.04, then dmucs not recognize the hosts

Install Caffe & TensorFlow with Python and CUDA (macOS)

Requirements

Python Installation

CUDA Installation

OpenCV Installation

Issues

Issue, make pycaffe:

Issue, import caffe:

Others

Issues

Issue, import tensorflow:

Issue, “Library not loaded: @rpath/libcublas.8.0.dylib”:

References

使用 Djinni 开发 Android, iOS 共享库

初见：Djinni 及其样例

下载 Djinni

编译 Djinni

使用 Djinni

编译样例

准备 GYP

从无到有：Hello Djinni

C++ 接口

定义接口描述文件

生成接口绑定代码

实现 C++ 接口

C++ 工程

iOS 工程

生成接口 Libraries 工程

Failed to use libfreenect form `sudo apt-get install freenect`

Failed to use OpenNI2 from `sudo apt-get install libopenni2-dev`

Failed to use OpenNI2 from `OpenNI-Linux-Arm-2.2.0.33.tar.bz2`

Issue when `cmake`

Issue when `make`

Issue: `loadavg` get wrong host on Ubuntu 14.04, then `dmucs` not recognize the hosts

Issue, `make pycaffe`:

Issue, `import caffe`:

Issue, `import tensorflow`: