iOS Dlib的集成

虽然 CocoadPods 有搜索到 pod，但是不可用。
我们采用下载源码，自己编译静态库的方式。

打包成静态库

由于dlib是用c++写的，虽然xcode知识Swift、OC、C、C++混编，但是对于这种三房库最好打成静态库。

接下来就大致介绍一下如何完成这个步骤。

• 前提条件

  1. X11(这个之前的mac是包含了的，但是现在需要自己下载安装)下载地址X11。
  2. Xcode
  3. cmake，如何之前玩过c或者c++这个应该安装过了。如果没安装过可以通过homebrew安装

• 步骤

  1. 将项目dlib clone 到本地。项目地址https://github.com/davisking/dlib.git
  2. 在终端里面打开刚才clone下来的项目，运行下面的命令

$ cd examples
$ mkdir build
$ cd build
$ cmake -G Xcode .. 
$ cmake --build . --config Release

如果没有出现异常，最后会出现

image.png
3.这一串命令最终会产生一个dlib_build目录，里面就是编译这个库的xcode项目。在这个项目里面可以根据你的需要设置这个SDK支持的架构和支持的平台。
image.png

集成到项目中

1.将刚才生产的.a和dlib的库文件拖到同一个文件夹。

image.png
image.png

2.将文件夹复制到项目中，目录结构如下

image.png
3.然后将libdlib.a加入到项目中，特别注意不要把库文件dlib目录下的文件拖进去。在文件中引用一下库文件。
image.png

4.设置header search

image.png

5.在需要用到的文件里引入头文件，将.m后缀改成.mm 后缀，编译成功

image.png ‘
6.编译成功

#######当集成openvc 和dlib报错 No viable conversion from 'const cv::Mat' to 'IplImage' (aka '_IplImage')时
替换cv_image.h文件如下

// Copyright (C) 2009  Davis E. King (davis@dlib.net)
// License: Boost Software License   See LICENSE.txt for the full license.
#ifndef DLIB_CvIMAGE_H_
#define DLIB_CvIMAGE_H_

#include <opencv2/core/core.hpp>
#include <opencv2/core/types_c.h>
#include "cv_image_abstract.h"
#include "../algs.h"
#include "../pixel.h"
#include "../matrix/matrix_mat.h"
#include "../image_processing/generic_image.h"

namespace dlib
{

    template <
        typename pixel_type
        >
    class cv_image
    {
    public:
        typedef pixel_type type;
        typedef default_memory_manager mem_manager_type;

        cv_image (const cv::Mat& img)
        {
            DLIB_CASSERT(img.depth() == cv::DataType<typename pixel_traits<pixel_type>::basic_pixel_type>::depth &&
                         img.channels() == pixel_traits<pixel_type>::num,
                         "The pixel type you gave doesn't match pixel used by the open cv Mat object."
                         << "\n\t img.depth():    " << img.depth()
                         << "\n\t img.cv::DataType<typename pixel_traits<pixel_type>::basic_pixel_type>::depth: "
                            << cv::DataType<typename pixel_traits<pixel_type>::basic_pixel_type>::depth
                         << "\n\t img.channels(): " << img.channels()
                         << "\n\t img.pixel_traits<pixel_type>::num: " << pixel_traits<pixel_type>::num
                         );
// Note, do NOT use CV_VERSION_MAJOR because in OpenCV 2 CV_VERSION_MAJOR actually held
// CV_VERSION_MINOR and instead they used CV_VERSION_EPOCH.  So for example, in OpenCV
// 2.4.9.1 CV_VERSION_MAJOR==4 and CV_VERSION_EPOCH==2.  However, CV_MAJOR_VERSION has always
// (seemingly) held the actual major version number, so we use that to test for the OpenCV major
// version.
#if CV_MAJOR_VERSION > 3 || (CV_MAJOR_VERSION == 3 && CV_SUBMINOR_VERSION >= 9)
            IplImage temp = cvIplImage(img);
#else
            IplImage temp = img;
#endif
            init(&temp);
        }

        cv_image (const IplImage img)
        {
            init(&img);
        }

        cv_image (const IplImage* img)
        {
            init(img);
        }

        cv_image() : _data(0), _widthStep(0), _nr(0), _nc(0) {}

        size_t size () const { return static_cast<size_t>(_nr*_nc); }

        inline pixel_type* operator[](const long row )
        {
            // make sure requires clause is not broken
            DLIB_ASSERT(0 <= row && row < nr(),
                "\tpixel_type* cv_image::operator[](row)"
                << "\n\t you have asked for an out of bounds row "
                << "\n\t row:  " << row
                << "\n\t nr(): " << nr()
                << "\n\t this:  " << this
                );

            return reinterpret_cast<pixel_type*>( _data + _widthStep*row);
        }

        inline const pixel_type* operator[](const long row ) const
        {
            // make sure requires clause is not broken
            DLIB_ASSERT(0 <= row && row < nr(),
                "\tconst pixel_type* cv_image::operator[](row)"
                << "\n\t you have asked for an out of bounds row "
                << "\n\t row:  " << row
                << "\n\t nr(): " << nr()
                << "\n\t this:  " << this
                );

            return reinterpret_cast<const pixel_type*>( _data + _widthStep*row);
        }

        inline const pixel_type& operator()(const long row, const long column) const
        {
          DLIB_ASSERT(0<= column && column < nc(),
              "\tcont pixel_type& cv_image::operator()(const long rown const long column)"
              << "\n\t you have asked for an out of bounds column "
              << "\n\t column: " << column
              << "\n\t nc(): " << nc()
              << "\n\t this:  " << this
              );

          return (*this)[row][column];
        }

        inline pixel_type& operator()(const long row, const long column)
        {
          DLIB_ASSERT(0<= column && column < nc(),
              "\tcont pixel_type& cv_image::operator()(const long rown const long column)"
              << "\n\t you have asked for an out of bounds column "
              << "\n\t column: " << column
              << "\n\t nc(): " << nc()
              << "\n\t this:  " << this
              );

          return (*this)[row][column];
        }

        long nr() const { return _nr; }
        long nc() const { return _nc; }
        long width_step() const { return _widthStep; }

    private:

        void init (const IplImage* img)
        {
            DLIB_CASSERT( img->dataOrder == 0, "Only interleaved color channels are supported with cv_image");
            DLIB_CASSERT((img->depth&0xFF)/8*img->nChannels == sizeof(pixel_type),
                         "The pixel type you gave doesn't match the size of pixel used by the open cv image struct");

            _data = img->imageData;
            _widthStep = img->widthStep;
            _nr = img->height;
            _nc = img->width;

        }

        char* _data;
        long _widthStep;
        long _nr;
        long _nc;
    };

// ----------------------------------------------------------------------------------------

    template <
        typename T
        >
    const matrix_op<op_array2d_to_mat<cv_image<T> > > mat (
        const cv_image<T>& m
    )
    {
        typedef op_array2d_to_mat<cv_image<T> > op;
        return matrix_op<op>(op(m));
    }

// ----------------------------------------------------------------------------------------

// Define the global functions that make cv_image a proper "generic image" according to
// ../image_processing/generic_image.h
    template <typename T>
    struct image_traits<cv_image<T> >
    {
        typedef T pixel_type;
    };

    template <typename T>
    inline long num_rows( const cv_image<T>& img) { return img.nr(); }
    template <typename T>
    inline long num_columns( const cv_image<T>& img) { return img.nc(); }

    template <typename T>
    inline void* image_data(
        cv_image<T>& img
    )
    {
        if (img.size() != 0)
            return &img[0][0];
        else
            return 0;
    }

    template <typename T>
    inline const void* image_data(
        const cv_image<T>& img
    )
    {
        if (img.size() != 0)
            return &img[0][0];
        else
            return 0;
    }

    template <typename T>
    inline long width_step(
        const cv_image<T>& img
    )
    {
        return img.width_step();
    }

// ----------------------------------------------------------------------------------------

}

#endif // DLIB_CvIMAGE_H_