opencv tutorials

wget https://github.com/opencv/opencv/archive/2.4.13.tar.gz
#wget https://github.com/Itseez/opencv/archive/3.1.0.tar.gz
#git clone https://github.com/opencv/opencv.git
cd opencv
mkdir build
cd build
cmake -D CMAKE_INSTALL_PREFIX=~/opencv_dir/ -D CMAKE_BUILD_TYPE=Release -D BUILD_EXAMPLES=ON ..
make -j4
make install

#git clone https://github.com/davisking/dlib.git
wget http://dlib.net/files/dlib-19.2.tar.bz2
cd dlib
mkdir build
cd build
cmake -D CMAKE_INSTALL_PREFIX=~/dlib_dir/ ..
make -j4
make install

export PKG_CONFIG_PATH=~/opencv_dir/lib/pkgconfig:$PKG_CONFIG_PATH
export PKG_CONFIG_PATH=~/dlib_dir/lib/pkgconfig:$PKG_CONFIG_PATH
export LD_LIBRARY_PATH=~/opencv_dir/lib:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=~/dlib_dir/lib:$LD_LIBRARY_PATH
export CMAKE_PREFIX_PATH=~/opencv_dir:$CMAKE_PREFIX_PATH
export CMAKE_PREFIX_PATH=~/dlib_dir:$CMAKE_PREFIX_PATH
g++ f.cpp -o f $(pkg-config --cflags --libs --static opencv) && ./f

import sys;sys.path.append('/home/user/opencv_dir/lib/python2.7/dist-packages');import cv2;import cv2.cv as cv;import numpy as np;

GraphCut vs GrabCut

（1）Graph Cut的目标和背景的模型是灰度直方图，Grab Cut取代为RGB三通道的混合高斯模型GMM；
（2）Graph Cut的能量最小化（分割）是一次达到的，而Grab Cut取代为一个不断进行分割估计和模型参数学习的交互迭代过程；
（3）Graph Cut需要用户指定目标和背景的一些种子点，但是Grab Cut只需要提供背景区域的像素集就可以了。也就是说你只需要框选目标，那么在方框外的像素全部当成背景，这时候就可以对GMM进行建模和完成良好的分割了。即Grab Cut允许不完全的标注（incomplete labelling）。

Blur

blur, boxFilter, bilateralFilter, GaussianBlur, medianBlur

DL

https://github.com/imistyrain/OpenDL