FFmpeg(GPU)-CUDA11.1-CUDNN8-OpenCV4.2(GPU)-Ubuntu18.04構建docker映象
阿新 • • 發佈：2022-02-25
FFmpeg(GPU)-CUDA11.1-CUDNN8-OpenCV4.2(GPU)-Ubuntu18.04構建docker映象
-
首先需要編譯安裝GPU-FFMPEG, 可參照ubuntu_FFMPEG編譯安裝
-
Video Code SDK 下載 需要登入，我下載是Video_Codec_SDK_11.1.5.zip ，解壓後將
Interface下的cuviddec.h和nvcuvid.h放到/usr/include -
依賴庫安裝
apt-get update -y # apt-get remove -y x264 libx264-dev apt-get install -y build-essential checkinstall cmake pkg-config yasm apt-get install -y git gfortran add-apt-repository -y "deb http://security.ubuntu.com/ubuntu xenial-security main" apt-get install -y libjpeg8-dev libjasper-dev libpng12-dev apt-get install -y libtiff5-dev apt-get install -y libavcodec-dev libavformat-dev libswscale-dev libdc1394-22-dev apt-get install -y libxine2-dev libv4l-dev apt-get install -y libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev apt-get install -y qt5-default libgtk2.0-dev libtbb-dev apt-get install -y libatlas-base-dev apt-get install -y libfaac-dev libmp3lame-dev libtheora-dev apt-get install -y libvorbis-dev libxvidcore-dev apt-get install -y libopencore-amrnb-dev libopencore-amrwb-dev apt-get install -y x264 v4l-utils
- 如x264安裝失敗，可編譯安裝，x264編譯安裝
git clone https://code.videolan.org/videolan/x264.git
cd x264
sudo ./configure --enable-shared --disable-asm
make && make install
- 其他依賴
apt-get install -y libprotobuf-dev protobuf-compiler apt-get install -y libgoogle-glog-dev libgflags-dev apt-get install -y libgphoto2-dev libeigen3-dev libhdf5-dev doxygen
- OPENGL 支援依賴庫
apt-get install mesa-common-dev libgtkglext1 libgtkglext1-dev freeglut3-dev
注意:可能出現freeglut3-dev無法安裝這裡我有安裝包直接:freeglut3-dev_2.8.1-3_amd64.deb,freeglut3_2.8.1-3_amd64.deb
dpkg -i <安裝包名稱即可.deb>
-
以下相關下載依賴包可以通過這裡連結下載, 提取碼：ip1a
-
Opencv和OpencvContrib下載
git clone https://github.com/opencv/opencv.git --branch=4.2.0 git clone https://github.com/opencv/opencv_contrib.git --branch=4.2.0
- 編譯安裝
cd opencv
mkdir build
cd build
cmake -D CMAKE_INSTALL_PREFIX=/usr/local/opencv-4.2.0 \
-D CMAKE_BUILD_TYPE=Debug \
-D OPENCV_EXTRA_MODULES_PATH=/opt/opencv_base/opencv_contrib/modules \
-D BUILD_opencv_python3=ON \
-D PYTHON_DEFAULT_EXECUTABLE=$(python3 -c "import sys; print(sys.executable)") \
-D BUILD_opencv_python2=OFF \
-D PYTHON3_EXECUTABLE=$(python -c "import sys; print(sys.executable)") \
-D PYTHON3_NUMPY_INCLUDE_DIRS=$(python3 -c "import numpy; print(numpy.get_include())") \
-D PYTHON3_PACKAGES_PATH=$(python3 -c "from distutils.sysconfig import get_python_lib; print(get_python_lib())") \
-D BUILD_opencv_hdf=OFF \
-D WITH_CUDA=ON \
-D WITH_OPENGL=ON \
-D WITH_OPENMP=ON \
-D WITH_GTK=ON \
-D WITH_VTK=ON -D WITH_TBB=ON \
-D WITH_GSTREAMER=ON \
-D WITH_CUDNN=ON \
-D WITH_CUBLAS=ON \
-D WITH_GTK_2_X=ON \
-D BUILD_EXAMPLES=ON \
-D OPENCV_ENABLE_NONFREE=ON \
-D WITH_FFMPEG=ON \
-D OPENCV_GENERATE_PKGCONFIG=ON \
-D WITH_NVCUVID=ON \
-D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-11.1 \
-D CUDA_ARCH_BIN=5.3,6.0,6.1,7.0,7.5 \
-D CUDA_ARCH_PTX=7.5 \
-D CUDA_nppicom_LIBRARY=stdc++ \
..
- 編譯註意事項：
1.cmake選項注意：
-D WITH_OPENGL=ON #開啟OPENGL，必須
-D WITH_GTK_2_X=ON #必須，否則OPENGL無法開啟
-D BUILD_EXAMPLES=ON #用於後續驗證GPU編解碼
-D WITH_NVCUVID=ON #用於安裝顯示卡編解碼相關庫 必須
2.docker中配置問題 無NVCUVID解決辦法: 需要將host(主機)中的libnvcuvid.so與libnvcuvid.so.1 libnvcuvid.so.4XX.XX.X(庫的實體)放入到docker的/usr/lib/x86_64-linux-gnu/路徑下，並在docker中建立軟連線，具體操作
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.450.80.02 /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/lib/x86_64-linux-gnu/libnvcuvid.so
sudo ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so /usr/lib/libnvcuvid.so
sudo ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/lib/libnvcuvid.so.1
# 將宿主機libnvcuvid.so.470.103.01 拷貝到 容器的 /usr/lib/x86_64-linux-gnu 下
docker cp /usr/lib/x86_64-linux-gnu/libnvcuvid.so.470.103.01 5d74c5af2bc8:/usr/lib/x86_64-linux-gnu
# 建立軟連線
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.470.103.01 /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/lib/x86_64-linux-gnu/libnvcuvid.so
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so /usr/lib/libnvcuvid.so
ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/lib/libnvcuvid.so.1
3.boostdesc_bgm.i相關下載：，將資源下載解壓拷貝到opencv_contrib/modules/xfeatures2d/src/目錄下，然後更改opencv_contrib/modules/xfeatures2d/cmake/下路徑
$ cd opencv_contrib/modules/xfeatures2d/src/
# 將 download_boostdesc.cmake 更換 https://raw.githubusercontent.com/opencv/opencv_3rdparty/${OPENCV_3RDPARTY_COMMIT}/ 為 file:///opt/opencv_base/opencv_contrib/modules/xfeatures2d/src/boostdesc_bgm
# download_vgg.cmake 也同理
4.face_landmark_model.dat，提前下載好face_landmark_model.dat，並放入映象中
vi opencv_contrib/modules/face/CMakeLists.txt
將https://raw.githubusercontent.com 替換為 file:///opt/opencv_base/face_landmark_model/
5.ippicv下載, 提期下載好ippicv_2019_lnx_intel64_general_20180723.tgz,並放入映象中
vi opencv/3rdparty/ippicv/ippicv.cmake
https://raw.githubusercontent.com/opencv/opencv_3rdparty/${IPPICV_COMMIT}/ippicv/ 修改為
/opt/opencv_base/ippicv/
- 參考
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D PYTHON_DEFAULT_EXECUTABLE=$(python3 -c "import sys; print(sys.executable)") \
-D PYTHON3_EXECUTABLE=$(python -c "import sys; print(sys.executable)") \
-D PYTHON3_NUMPY_INCLUDE_DIRS=$(python -c "import numpy; print(numpy.get_include())") \
-D PYTHON3_PACKAGES_PATH=$(python -c "from distutils.sysconfig import get_python_lib; print(get_python_lib())") \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D WITH_TBB=ON \
-D WITH_V4L=ON \
-D WITH_OPENGL=ON \
-D WITH_CUDA=ON \
-D WITH_CUDNN=ON \
-D ENABLE_FAST_MATH=1 \
-D CUDA_FAST_MATH=1 \
-D CUDA_NVCC_FLAGS="-D_FORCE_INLINES" \
-D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-11.1 \
-D WITH_NVCUVID=ON \
-D CUDA_ARCH_BIN=5.3,6.0,6.1,7.0,7.5 \
-D CUDA_ARCH_PTX=7.5 \
-D WITH_CUBLAS=1 \
-D OPENCV_EXTRA_MODULES_PATH=/opt/opencv_base/opencv_contrib/modules \
-D CUDA_nppicom_LIBRARY=stdc++ \
-D WITH_OPENMP=ON \
-D CUDA_nppicom_LIBRARY=stdc++ \
..
- cmake結果
-- Disabled: hdf world
-- Disabled by dependency: -
-- Unavailable: cnn_3dobj cvv java js matlab ovis python2 viz xfeatures2d
-- Applications: tests perf_tests examples apps
-- Documentation: NO
-- Non-free algorithms: YES
--
-- GUI:
-- GTK+: YES (ver 2.24.32)
-- GThread : YES (ver 2.56.4)
-- GtkGlExt: YES (ver 1.2.0)
-- OpenGL support: YES (/usr/lib/x86_64-linux-gnu/libGL.so /usr/lib/x86_64-linux-gnu/libGLU.so)
-- VTK support: NO
--
-- Media I/O:
-- ZLib: /usr/lib/x86_64-linux-gnu/libz.so (ver 1.2.11)
-- JPEG: /usr/lib/x86_64-linux-gnu/libjpeg.so (ver 80)
-- WEBP: build (ver encoder: 0x020e)
-- PNG: /usr/lib/x86_64-linux-gnu/libpng.so (ver 1.6.34)
-- TIFF: /usr/lib/x86_64-linux-gnu/libtiff.so (ver 42 / 4.0.9)
-- JPEG 2000: /usr/lib/x86_64-linux-gnu/libjasper.so (ver 1.900.1)
-- OpenEXR: build (ver 2.3.0)
-- HDR: YES
-- SUNRASTER: YES
-- PXM: YES
-- PFM: YES
--
-- Video I/O:
-- DC1394: YES (2.2.5)
-- FFMPEG: YES
-- avcodec: YES (57.107.100)
-- avformat: YES (57.83.100)
-- avutil: YES (55.78.100)
-- swscale: YES (4.8.100)
-- avresample: NO
-- GStreamer: YES (1.14.5)
-- v4l/v4l2: YES (linux/videodev2.h)
--
-- Parallel framework: TBB (ver 2017.0 interface 9107)
--
-- Trace: YES (with Intel ITT)
--
-- Other third-party libraries:
-- Intel IPP: 2019.0.0 Gold [2019.0.0]
-- at: /opt/opencv_base/opencv/build/3rdparty/ippicv/ippicv_lnx/icv
-- Intel IPP IW: sources (2019.0.0)
-- at: /opt/opencv_base/opencv/build/3rdparty/ippicv/ippicv_lnx/iw
-- Lapack: NO
-- Eigen: YES (ver 3.3.4)
-- Custom HAL: NO
-- Protobuf: build (3.5.1)
--
-- NVIDIA CUDA: YES (ver 11.1, CUFFT CUBLAS NVCUVID)
-- NVIDIA GPU arch: 53 60 61 70 75
-- NVIDIA PTX archs: 75
--
-- cuDNN: NO
--
-- OpenCL: YES (no extra features)
-- Include path: /opt/opencv_base/opencv/3rdparty/include/opencl/1.2
-- Link libraries: Dynamic load
--
-- Python 3:
-- Interpreter: /usr/bin/python (ver 3.6.9)
-- Libraries: /usr/lib/x86_64-linux-gnu/libpython3.6m.so (ver 3.6.9)
-- numpy: /usr/local/lib/python3.6/dist-packages/numpy/core/include (ver 1.16.1)
-- install path: /usr/lib/python3/dist-packages/cv2/python-3.6
--
-- Python (for build): /usr/bin/python3
--
-- Java:
-- ant: NO
-- JNI: NO
-- Java wrappers: NO
-- Java tests: NO
--
-- Install to: /usr/local/opencv-4.2.0
-- -----------------------------------------------------------------
--
-- Configuring done
-- Generating done
-- Build files have been written to: /opt/opencv_base/opencv/build
- make
make -j8
編譯報錯:feature2d/test/test_detectors_regression.impl.hpp
# 報錯feature2d/test/test_detectors_regression.impl.hpp
# 操作:將opencv/modules/features2d拷貝到，build下
# cd目錄下/opt/opencv_base/opencv/build
cp -r ../../opencv/modules/features2d/ .
編譯報錯:/opt/opencv_base/opencv/samples/gpu/surf_keypoint_matcher.cpp fatal error: opencv2/xfeatures2d/cuda.hpp: No such file or directory include "opencv2/xfeatures2d/cuda.hpp"
vi /opt/opencv_base/opencv/samples/gpu/surf_keypoint_matcher.cpp
將#include "opencv2/xfeatures2d/cuda.hpp" 改為 #include "/opt/opencv_base/opencv_contrib/modules/xfeatures2d/include/opencv2/xfeatures2d/cuda.hpp"
編譯報錯SURF_CUDA::SURF_CUDA解決方法1:/opt/opencv_base/opencv/samples/gpu/surf_keypoint_matcher.cpp:53: undefined reference to cv::cuda::SURF_CUDA::SURF_CUDA()參考 親測有效
vi /opt/opencv_base/opencv/build/samples/gpu/CMakeFiles/example_gpu_surf_keypoint_matcher.dir/link.txt
# 將 CMakeFiles/example_gpu_surf_keypoint_matcher.dir/surf_keypoint_matcher.cpp.o 替換成如下
--as-needed CMakeFiles/example_gpu_surf_keypoint_matcher.dir/surf_keypoint_matcher.cpp.o ../../modules/xfeatures2d/CMakeFiles/opencv_xfeatures2d.dir/src/surf.cuda.cpp.o ../../modules/xfeatures2d/CMakeFiles/cuda_compile_1.dir/src/cuda/cuda_compile_1_generated_surf.cu.o
編譯報錯SURF_CUDA::SURF_CUDA解決方法2:/opt/opencv_base/opencv/samples/gpu/surf_keypoint_matcher.cpp:53: undefined reference to cv::cuda::SURF_CUDA::SURF_CUDA()參考 沒有親測
# 找到opencv下CMAKELists.txt更改
OCV_OPTION(OPENCV_ENABLE_NONFREE "Enable non-free algorithms" OFF)
# OFF 改為 ON
OCV_OPTION(OPENCV_ENABLE_NONFREE "Enable non-free algorithms" ON)
# 清理
make clean
# 重新cmake
- 安裝
make install
- OpenCV環境變數新增 (到這裡了)
vi /etc/ld.so.conf.d/opencv.conf
/usr/local/opencv-4.2.0/lib
# 更新
ldconfig
- 配置bash
vi /etc/bash.bashrc
//在末尾新增
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/opencv-4.2.0/lib/pkgconfig
export PKG_CONFIG_PATH
# 儲存退出，然後執行如下命令使得配置生效
source /etc/bash.bashrc
sudo updatedb
- 配置本地bash
vi ~/.bashrc
# 在檔案末尾增加以下內容
export PKG_CONFIG_PATH=/usr/local/opencv-4.2.0/lib/pkgconfig
export LD_LIBRARY_PATH=/usr/local/opencv-4.2.0/lib
# 更新~/.bashrc
source ~/.bashrc
- 檢視openCV版本
pkg-config opencv4 --modversion
- 建立opencv4.pc
vi /usr/local/lib/pkgconfig/opencv4.pc
# Package Information for pkg-config
prefix=/usr/local
exec_prefix=${prefix}
libdir=${exec_prefix}/lib
includedir_old=${prefix}/include/opencv4/opencv
includedir_new=${prefix}/include/opencv4
Name: OpenCV
Description: Open Source Computer Vision Library
Version: 4.2.0
Libs: -L${exec_prefix}/lib -lopencv_gapi -lopencv_stitching -lopencv_aruco -lopencv_bgsegm -lopencv_bioinspired -lopencv_ccalib -lopencv_dnn_objdetect -lopencv_dpm -lopencv_face -lopencv_freetype -lopencv_fuzzy -lopencv_hfs -lopencv_img_hash -lopencv_line_descriptor -lopencv_quality -lopencv_reg -lopencv_rgbd -lopencv_saliency -lopencv_stereo -lopencv_structured_light -lopencv_phase_unwrapping -lopencv_superres -lopencv_optflow -lopencv_surface_matching -lopencv_tracking -lopencv_datasets -lopencv_text -lopencv_dnn -lopencv_plot -lopencv_videostab -lopencv_video -lopencv_xfeatures2d -lopencv_shape -lopencv_ml -lopencv_ximgproc -lopencv_xobjdetect -lopencv_objdetect -lopencv_calib3d -lopencv_features2d -lopencv_highgui -lopencv_videoio -lopencv_imgcodecs -lopencv_flann -lopencv_xphoto -lopencv_photo -lopencv_imgproc -lopencv_core
Libs.private: -ldl -lm -lpthread -lrt
Cflags: -I${includedir_old} -I${includedir_new}
- pkg-config測試
pkg-config --cflags --libs opencv4
-
C++指令碼測試
opencv_cuda.cpp
#include <iostream>
#include "opencv2/opencv_modules.hpp"
#if defined(HAVE_OPENCV_CUDACODEC)
#include <string>
#include <vector>
#include <algorithm>
#include <numeric>
#include <opencv2/core.hpp>
#include <opencv2/core/opengl.hpp>
#include <opencv2/cudacodec.hpp>
#include <opencv2/highgui.hpp>
int main(int argc, const char* argv[])
{
if (argc != 2)
return -1;
std::cout << "1. parse script query" << std::endl;
const std::string fname(argv[1]);
// cv::namedWindow("CPU", cv::WINDOW_NORMAL);
// cv::namedWindow("GPU", cv::WINDOW_OPENGL);
// http://cn.voidcc.com/question/p-ehnwpjbe-wb.html
// cv::cuda::setGlDevice();
std::cout << "2. set gl device" << std::endl;
//cv::Mat frame;
//cv::VideoCapture reader(fname);
cv::cuda::GpuMat d_frame;
cv::Ptr<cv::cudacodec::VideoReader> d_reader = cv::cudacodec::createVideoReader(fname);
std::cout << "3.create video" << std::endl;
cv::TickMeter tm;
std::vector<double> cpu_times;
std::vector<double> gpu_times;
int gpu_frame_count=0, cpu_frame_count=0;
/*
for (;;)
{
tm.reset(); tm.start();
if (!reader.read(frame))
break;
tm.stop();
cpu_times.push_back(tm.getTimeMilli());
cpu_frame_count++;
cv::imshow("CPU", frame);
if (cv::waitKey(3) > 0)
break;
}
*/
for (;;)
{
tm.reset(); tm.start();
if (!d_reader->nextFrame(d_frame))
break;
tm.stop();
gpu_times.push_back(tm.getTimeMilli());
gpu_frame_count++;
// cv::imshow("GPU", d_frame);
if (cv::waitKey(3) > 0)
break;
}
if (!cpu_times.empty() || !gpu_times.empty())
{
std::cout << std::endl << "Results:" << std::endl;
//std::sort(cpu_times.begin(), cpu_times.end());
std::sort(gpu_times.begin(), gpu_times.end());
//double cpu_avg = std::accumulate(cpu_times.begin(), cpu_times.end(), 0.0) / cpu_times.size();
double gpu_avg = std::accumulate(gpu_times.begin(), gpu_times.end(), 0.0) / gpu_times.size();
//std::cout << "CPU : Avg : " << cpu_avg << " ms FPS : " << 1000.0 / cpu_avg << " Frames " << cpu_frame_count << std::endl;
std::cout << "GPU : Avg : " << gpu_avg << " ms FPS : " << 1000.0 / gpu_avg << " Frames " << gpu_frame_count << std::endl;
}
return 0;
}
#else
int main()
{
std::cout << "OpenCV was built without CUDA Video decoding support\n" << std::endl;
return 0;
}
#endif
註釋掉cv::namedWindow和cv::imshow部分 因為是在docker映象執行
註釋掉cv::cuda::setGlDevice 可在這裡找到解釋說明
Makefile編寫
opencv_cuda.o:opencv_cuda.cpp
g++ -std=c++14 -g -o main.out opencv_cuda.cpp `pkg-config opencv4 --cflags --libs` \
-I/usr/local/opencv-4.2.0/include/opencv4/opencv2 \
-I/usr/local/cuda/include \
-L/usr/local/cuda/lib64 \
-I/usr/include/eigen3 \
-L/usr/lib/x86_64-linux-gnu -lcuda -ldl -lnvcuvid
clean:
rm *.o main.out
-
執行
make -
測試
./main.out rtsp://admin:[email protected]:554
- 參考
- https://note.youdao.com/ynoteshare/index.html?id=700052b0a49301059a34f20a00a830ca&type=note&_time=1645577353739
- https://w7536214.github.io/2020/07/28/ffmpeg_note_1/
- https://my.oschina.net/u/4391831/blog/3312094
- https://www.jb51.net/article/191623.htm
- https://blog.csdn.net/GungnirsPledge/article/details/108597474