people_detect_pi

1. Build and install OpenCV (3.3.1)

reference:Optimizing OpenCV on the Raspberry Pi - PyImageSearch

# Build tools
sudo apt install build-essential cmake pkg-config
# Image read
sudo apt install libjpeg-dev libtiff5-dev libjasper-dev libpng-dev
# Video read
sudo apt install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev
sudo apt install libxvidcore-dev libx264-dev
# GUI
sudo apt install libgtk2.0-dev libgtk-3-dev
sudo apt install libcanberra-gtk*
# Matrix operation optimization
sudo apt install libatlas-base-dev gfortran
# Python header
sudo apt install python2.7-dev python3-dev pylint pylint3
# Increase the swap partition capacity
sudo nano /etc/dphys-swapfile
# Set CONF_SWAPSIZE to 1024
# Restart swap service
sudo /etc/init.d/dphys-swapfile stop
sudo /etc/init.d/dphys-swapfile start
# Build and install Ceres Solver
sudo apt install libeigen3-dev libgflags-dev libgoogle-glog-dev
sudo apt install libsuitesparse-dev
cd ~
# git clone https://ceres-solver.googlesource.com/ceres-solver
git clone https://github.com/ceres-solver/ceres-solver
git checkout <new version>
cd ceres-solver
mkdir build && cd build
cmake ..
make -j4 # several hours
make test # one hour
sudo make install
# Make opencv build dir
cd ~
mkdir opencv-3.3.1
cd opencv-3.3.1
# Clone opencv
git clone https://github.com/opencv/opencv.git
cd opencv
git checkout 3.3.1
cd ..
# Clone opencv_contrib
git clone https://github.com/opencv/opencv_contrib.git
cd opencv_contrib
git checkout 3.3.1
cd ..
# Configure cmake compilation parameters:

# BUILD_EXAMPLES
# BUILD_opencv_world BUILD_opencv_contrib_world
# BUILD_opencv_python2 BUILD_opencv_python3
# BUILD_TBB WITH_TBB
# ENABLE_NEON ENABLE_VFPV3
# INSTALL_CREATE_DISTRIB
# INSTALL_C_EXAMPLES
# INSTALL_TESTS
# OPENCV_EXTRA_MODULES_PATH
mkdir build
cd build
cmake \
    -D BUILD_EXAMPLES=ON \
    -D BUILD_opencv_world=ON \
    -D BUILD_opencv_contrib_world=ON \
    -D BUILD_opencv_python2=ON \
    -D BUILD_opencv_python3=ON \
    -D BUILD_TBB=ON \
    -D WITH_TBB=ON \
    -D ENABLE_NEON=ON \
    -D ENABLE_VFPV3=ON \
    -D INSTALL_CREATE_DISTRIB=ON \
    -D INSTALL_C_EXAMPLES=ON \
    -D INSTALL_PYTHON_EXAMPLES=ON \
    -D INSTALL_TESTS=ON \
    -D OPENCV_EXTRA_MODULES_PATH=../opencv_contrib/modules  \
    ../opencv 

make -j2
sudo make install
# Restore swap partition size
sudo nano /etc/dphys-swapfile
# Set CONF_SWAPSIZE to 100
# Restart swap service
sudo /etc/init.d/dphys-swapfile stop
sudo /etc/init.d/dphys-swapfile start
# In order to prevent:
#   Error retrieving accessibility bus address: org.freedesktop.DBus.Error.ServiceUnknown: The name org.a11y.Bus was not provided by any .service files
# We need to install at-spi2-core
sudo apt install at-spi2-core

2. Build and install raspicam

cd ~
git clone https://github.com/cedricve/raspicam
cd raspicam
mkdir build
cd build
cmake ..
make -j4
sudo make install
sudo ldconfig

3. Build and install matio

# Install hdf5
sudo apt install libhdf5-dev
cd ~
git clone git://git.code.sf.net/p/matio/matio
cd matio
git submodule update --init  # for datasets used in unit tests
./autogen.sh
# Create a fake directory so matio can recognize hdf5
mkdir fake_hdf5_dir
cd fake_hdf5_dir
ln -s /usr/include/hdf5/serial/ ./include
ln -s /usr/lib/arm-linux-gnueabihf/hdf5/serial/ ./lib
cd ..
./configure --enable-mat73=yes --with-hdf5=/home/pi/matio/fake_hdf5_dir/ --with-default-api-version=v110 --with-default-file-ver=7.3
make -j4
make check  # test
sudo make install
sudo ldconfig

4. Use checkinstall to package and copy to Raspberry Pi computing module and install libopencv libmatio libraspicam

5. Handle dependent dynamic link libraries

sudo apt install libtiff5 libjasper1 libavcodec57 libavformat57 libswscale4
sudo apt install libhdf5-100
sudo apt install libgtk-3-0

[note] this part has been done in the package, if you compile, follow these steps:

6. Build source code

• Copy the people-detect-pi-src.zip decompressed file to the Raspberry Pi home directory
• Copy AcfHSMy18Detector.mat to the Raspberry Pi home directory

cd ~
cd people-detect-pi-src
mkdir people-detect-pi-build
cmake ../
make

7. Run

• Copy the ACF_HS_Detect file under people-detect-pi-src / people-detect-pi-build to the Raspberry Pi home directory
• Copy human.sh to the Raspberry Pi desktop
• run human.sh