基于Ubuntu20.04 x86 Docker（可调用CUDA及CUDNN）环境下编译支持Gstreamer调用Nvidia显卡视频硬件编解码的opencv-python

目录

前言

一、编译Gstreamer

二、编译opencv-python

---

前言

  本篇博客旨在帮助读者快速搭建linux x86系统架构下Python3调用Nvidia显卡进行并行计算（CUDA、CUDNN）及视频硬件编解码（NVenc、NVdec）的基础环境。默认系统中Docker环境及相关配置已全部设置完毕不再赘述，拉取nvidia官方提供的ubuntu20.04+cuda.x+cudnn.x基础镜像，使用该镜像创建容器并进入容器。

sudo docker pull nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04 # 拉取镜像
sudo docker run -it -d --gpus all --name build_env -e  NVIDIA_DRIVER_CAPABILITIES=compute,utility,video -v /home/user/Downloads:/home/sources nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04 /bin/bash # 创建容器
sudo docker start build_env # 启动容器
sudo docker attach build_env # 进入容器

一、编译Gstreamer

 更新并升级软件包

apt update && apt upgrade

下载并安装gstreamer依赖

apt install build-essential dpkg-dev flex bison autotools-dev automake liborc-dev autopoint libtool gtk-doc-tools python3-pip libmount-dev git wget bison flex ninja-build  libglib2.0-dev nasm
apt install valgrind libgirepository1.0-dev libcap-dev libgtk-3-dev libunwind-dev clzip gobject-introspection libdw-dev
apt install libxv-dev libasound2-dev libtheora-dev libogg-dev libvorbis-dev
apt install libbz2-dev libv4l-dev libvpx-dev libjack-jackd2-dev libsoup2.4-dev libpulse-dev
apt install faad libfaad-dev libfaac-dev libx264-dev libmad0-dev

下载gstreamer源码到指定文件夹，并编译安装 

git clone git@gitlab.freedesktop.org:gstreamer/gstreamer.git  # 拉取远程仓库源代码

退出容器再重新进入容器，防止pip3将meson安装到/usr/.local目录下。meson编译时如果有依赖自动下载出错，自己用浏览器下载后解压到subprojects下并修改为对应名称即可。

python3 -m pip install pip --upgrade
pip3 install meson
cd /home/resources/gstreamer
meson setup --prefix /usr build -Dgst-plugins-bad:nvcodec=enabled # 编译  
sudo ninja -C build install # 安装


gst-inspect-1.0  --version #查看是否安装成功
gst-inspect-1.0 -a | grep nvh264

二、编译opencv-python

安装opencv依赖库

apt install build-essential cmake git python3-dev python3-numpy libavcodec-dev libavformat-dev libswscale-dev libgtk-3-dev libgtk2.0-dev libcanberra-gtk-module libpng-dev libjpeg-dev libopenexr-dev libtiff-dev libwebp-dev libopencv-dev x264 libx264-dev libssl-dev ffmpeg

编译安装opencv-python 

python3 -m pip install opencv-python-4.6.0.66.tar.gz