跨平台 C++ 通用库，Dlib 18.15 发布

Dlib是一个使用现代C++技术编写的跨平台的通用库，遵守Boost Software licence.

主要特点如下：

1.完善的文档：每个类每个函数都有详细的文档，并且提供了大量的示例代码，如果你发现文档描述不清晰或者没有文档，告诉作者，作者会立刻添加。

2.可移植代码：代码符合ISO C++标准，不需要第三方库支持，支持win32、Linux、Mac OS X、Solaris、HPUX、BSDs 和 POSIX 系统

3.线程支持：提供简单的可移植的线程API

4.网络支持：提供简单的可移植的Socket API和一个简单的Http服务器

5.图形用户界面：提供线程安全的GUI API

6.数值算法：矩阵、大整数、随机数运算等

7.机器学习算法：

8.图形模型算法：

9.图像处理：支持读写Windows BMP文件，不同类型色彩转换

10.数据压缩和完整性算法：CRC32、Md5、不同形式的PPM算法

11.测试：线程安全的日志类和模块化的单元测试框架以及各种测试assert支持

12.一般工具：XML解析、内存管理、类型安全的big/little endian转换、序列化支持和容器类Dlib 18.15 发布，此版本更新内容如下：

新特性
   - Added a number of tools for working with 3D data:
      - Added the perspective_window which is a tool for displaying 3D point clouds.
      - Added camera_transform.  It performs the 3D to 2D mapping needed to visualize 3D
        data.
      - Added point_transform_affine3d as well as functions for creating such transforms:
        rotate_around_x(), rotate_around_y(), rotate_around_z(), and translate_point().
   - Added draw_solid_circle() for drawing on images.
   - Added get_best_hough_point() to the hough_transform.
   - Thanks to Jack Culpepper, the python API for object detection now outputs detection
     confidences.  
   - Added lspi, an implementation of the least-squares policy iteration algorithm.
      
非向后兼容特性
   - The shape_predictor and shape_predictor_trainer had a non-optimal behavior when used
     with objects that have non-square bounding boxes. This has been fixed but will cause
     models that were trained with the previous version of dlib to not work as accurately if
     they used non-square boxes. So you might have to retrain your models when updating dlib.

Bug 修复
   - Fixed a bug which prevented add_image_rotations() from compiling.

其他改进
   - The imglab tool now allows the user to click and drag annotations around by holding
     shift and right clicking.