Opencv 之 DNN 与 CUDA综述

明月醉窗台

已于 2024-07-23 08:56:34 修改

阅读量1.4k

点赞数 2

分类专栏： CUDA - Opencv 文章标签： opencv dnn 计算机视觉

于 2023-01-27 21:08:25 首次发布

本文链接：https://blog.csdn.net/yohnyang/article/details/128772077

版权

CUDA - Opencv 专栏收录该内容

27 篇文章 2 订阅 ¥19.90 ¥99.00

订阅专栏

超级会员免费看

本文档详细介绍了如何结合CUDA和OpenCV处理复杂图像数据，包括CUDA的基础知识、线程管理和内存结构，以及OpenCV的DNN模块在深度学习应用中的实践，如分类、检测和分割。此外，还探讨了CUDA在OpenCV中的加速作用，提供了丰富的学习资源和链接。

摘要由CSDN通过智能技术生成

专栏目录

内容越来越多，所以设置付费了，每一篇都是认认真真总结的一些理论、原理或者技巧
欢迎同道中人一起交流成长

#可通过git下载拉取
git clone https://github.com/opencv/opencv_extra.git

开源Opencv+contrib - 455：https://gitcode.net/openmodel/opencv
- 深度学习是当今计算机视觉中最受欢迎和增长最快的领域。从 OpenCV 3.1 开始，库中有 DNN 模块，该模块通过深度网络实现正向传递（推理），并使用一些流行的深度学习框架（如 Caffe）进行预训练。在OpenCV 3.3中，该模块已从opencv_contrib存储库提升到主存储库（https://github.com/opencv/opencv/tree/master/modules/dnn），并已显着加速。
- 该模块没有任何额外的依赖关系，除了libprotobuf，libprotobuf现在包含在OpenCV中。

1.DNN模块

opencv中的DNN模块对接深度学习框架的一些API、C++复现过来的深度学习模型（目标检测、分割）
官方介绍：https://docs.opencv.org/4.x/d2/d58/tutorial_table_of_content_dnn.html

#ifndef OPENCV_DNN_HPP
#define OPENCV_DNN_HPP

// This is an umbrella header to include into you project.
// We are free to change headers layout in dnn subfolder, so please include
// this header for future compatibility


/** @defgroup dnn Deep Neural Network module
  @{
    This module contains:
        - API for new layers creation, layers are building bricks of neural networks;
        - set of built-in most-useful Layers;
        - API to construct and modify comprehensive neural networks from layers;
        - functionality for loading serialized networks models from different frameworks.

    Functionality of this module is designed only for forward pass computations (i.e. network testing).
    A network training is in principle not supported.
  @}
*/
/** @example samples/dnn/classification.cpp
Check @ref tutorial_dnn_googlenet "the corresponding tutorial" for more details
*/
/** @example samples/dnn/colorization.cpp
*/
/** @example samples/dnn/object_detection.cpp
Check @ref tutorial_dnn_yolo "the corresponding tutorial" for more details
*/
/** @example samples/dnn/openpose.cpp
*/
/** @example samples/dnn/segmentation.cpp
*/
/** @example samples/dnn/text_detection.cpp
*/
#include <opencv2/dnn/dnn.hpp>

#endif /* OPENCV_DNN_HPP */

2. CUDA模块

CUDA模块即Opencv中的GPU模块，用于视觉算法实现过程中的加速
官方介绍：https://docs.opencv.org/4.x/da/d2c/tutorial_table_of_content_gpu.html

cuda的使用在目录 .\opencv-contrib\include\opencv2\dnn\dnn.cpp中可以找到

//其中包括
    /**
     * @brief Enum of computation backends supported by layers.
     * @see Net::setPreferableBackend
     */
    enum Backend
    {
        //! DNN_BACKEND_DEFAULT equals to DNN_BACKEND_INFERENCE_ENGINE if
        //! OpenCV is built with Intel's Inference Engine library or
        //! DNN_BACKEND_OPENCV otherwise.
        DNN_BACKEND_DEFAULT = 0,
        DNN_BACKEND_HALIDE,
        DNN_BACKEND_INFERENCE_ENGINE,            //!< Intel's Inference Engine computational backend
                                                 //!< @sa setInferenceEngineBackendType
        DNN_BACKEND_OPENCV,
        DNN_BACKEND_VKCOM,
        DNN_BACKEND_CUDA,
        DNN_BACKEND_WEBNN,
#ifdef __OPENCV_BUILD
        DNN_BACKEND_INFERENCE_ENGINE_NGRAPH = 1000000,     // internal - use DNN_BACKEND_INFERENCE_ENGINE + setInferenceEngineBackendType()
        DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019,      // internal - use DNN_BACKEND_INFERENCE_ENGINE + setInferenceEngineBackendType()
#endif
    };

    /**
     * @brief Enum of target devices for computations.
     * @see Net::setPreferableTarget
     */
    enum Target
    {
        DNN_TARGET_CPU = 0,
        DNN_TARGET_OPENCL,
        DNN_TARGET_OPENCL_FP16,
        DNN_TARGET_MYRIAD,
        DNN_TARGET_VULKAN,
        DNN_TARGET_FPGA,  //!< FPGA device with CPU fallbacks using Inference Engine's Heterogeneous plugin.
        DNN_TARGET_CUDA,
        DNN_TARGET_CUDA_FP16,
        DNN_TARGET_HDDL
    };