HackMD
opencv3.3.0 with deep learning
https://github.com/opencv/opencv/wiki/Deep-Learning-in-OpenCV
environment: lenovo x220 laptop
OS: ubuntu16.04.2
build opencv-3.3.0 on raspberry pi3
Install dependency
sudo apt-get update
Install dependency
sudo apt-get install build-essential cmake pkg-config libjpeg8-dev libtiff5-dev libjasper-dev libpng12-dev libavcodec-dev libavformat-dev libswscale-dev libv4l-dev libxvidcore-dev libx264-dev libgtk-3-dev libatlas-base-dev gfortran
Install python3.5
sudo apt-get install python3.5-dev
opencv-3.3.0 build from source
unzip opencv-3.3.0
cd opencv-3.3.0
mkdir build
cd build
CMake will look for CMakeLists.txt and generate Makefile. CMake confoguration (now we are in build/)
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D INSTALL_PYTHON_EXAMPLES=ON \
-D INSTALL_C_EXAMPLES=OFF \
-D PYTHON_EXECUTABLE=/usr/bin/python3.5 \
-D BUILD_EXAMPLES=ON ..\
-D WITH_IPP=ON ..
make -j4
sudo make install
mostly we utilize pkg-config to compile openCV code
configure OpenCV. open the opencv.conf file with the command
sudo gedit /etc/ld.so.conf.d/opencv.conf
/usr/local/lib
This set the .so (dynamic link library) searching path.
it will search opencv___.so from the path
Remember we alread copy .so to /usr/local/lib when typing sudo make install in previous stage.
configure the library
sudo ldconfig
now, edit ~/.bashrc
this would auto load terminal environment for pkg-config
sudo vim ~/.bashrc
add at the bottom
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig
export PKG_CONFIG_PATH
export PYTHONPATH=/usr/local/bin/python3.5:$PYTHONPATH
then save it.
test
inside opencv-3.3.0/build/bin/
has executable dnn-…
for dl
you need to collect 3 file
.cafffemodle
.prototxt
weight
for caffe net(zoo)
for FCN object detection
caffenet
FCN object detection(people)
mobileNetSSD
./example_dnn-ssd_mobilenet_object_detection -video=/dev/video1 -model=MobileNetSSD_train.caffemodel -proto=MobileNetSSD_300x300.prototxt
and .zip file include executable and .caffemodel , proto.txt for someone would like to try it. size about 600MB
https://drive.google.com/file/d/0BwY-lpO6tzxHRHNCdlRKczIzaEU/view?usp=sharing
only contain caffemodel and proto
https://drive.google.com/file/d/0BwY-lpO6tzxHVUZqN1pXc0xZdDQ/view?usp=sharing
reference:
http://www.coldvision.io/2016/07/29/image-classification-deep-learning-cnn-caffe-opencv-3-x-cuda/
https://github.com/opencv/opencv_contrib/issues/830
https://github.com/opencv/opencv_contrib/issues/941
https://github.com/opencv/opencv/issues/9322
