从头制作 Jetson Nano 系统：DeepStream SDK 快速入门

官网地址：https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Quickstart.html

NVIDIA® DeepStream Software Development Kit (SDK) is an accelerated AI framework to build intelligent video analytics (IVA) pipelines. DeepStream runs on NVIDIA® T4, NVIDIA® Ampere and platforms such as NVIDIA® Jetson™ Nano, NVIDIA® Jetson AGX Xavier™, NVIDIA® Jetson Xavier NX™, NVIDIA®Jetson™ TX1 and TX2.

1. Jetson Setup

在安装 DeepStream SDK 之前，我们先来看一下如何为 Jetson 做些准备工作。

1.1 安装 Jetson SDK 组件

下载 NVIDIA SDK Manager，地址为 https://developer.nvidia.com/embedded/jetpack。 你接下来会基于这套系统安装  JetPack 4.5.1 GA (对应 L4T 32.5.1 发发布版本) 。

• NVIDIA SDK Manager 是一个图形化的应用程序，负责烧录和安装 Jetpack 软件包。

• 烧录过程持续大约 10-30 分钟，这取决于你使用的电脑主机系统

这里我得补充说明一下。对于 Jetson 来讲，上述过程包括：

1. 下载 Jetson 系统镜像。这个要根据你的硬件型号选择对应的镜像即可。我用的是 Nano 2G 版本，下载后得到文件 jetson-nano-2gb-sd-card-image。我在 Win10 下给它加上扩展名 tar，解压出镜像文件 sd-blob.img。
2. 用软件 balenaEtcher 烧录 SD 卡，建议选择 64G 容量的卡，不然的话你会发现开发过程中容量不够用。

上述过程我会专门写文详细介绍，目前先大致了解这么多。

官方提示：如果你是用 Jetson Nano 或 Jetson Xavier NX 开发套件，你可以从 https://developer.nvidia.com/embedded/jetpack 下载 SD卡镜像文件。这套镜像包含了 CUDA、TensorRT 和 cuDNN。

 1.2 安装依赖项

输入下面的命令行，安装依赖项：

$ sudo apt install \
libssl1.0.0 \
libgstreamer1.0-0 \
gstreamer1.0-tools \
gstreamer1.0-plugins-good \
gstreamer1.0-plugins-bad \
gstreamer1.0-plugins-ugly \
gstreamer1.0-libav \
libgstrtspserver-1.0-0 \
libjansson4=2.11-1

安装 librdkafka (为消息代理启用Kafka协议适配器)

1. 从 GitHub 克隆 librdkafka 库：

   $ git clone https://github.com/edenhill/librdkafka.git

    

2. 配置并编译构建库：

   $ cd librdkafka
   $ git reset --hard 7101c2310341ab3f4675fc565f64f0967e135a6a
   $ ./configure
   $ make
   $ sudo make install

    

3. 把生成的库文件复制到 deepstream 目录：

   $ sudo mkdir -p /opt/nvidia/deepstream/deepstream-5.1/lib
   $ sudo cp /usr/local/lib/librdkafka* /opt/nvidia/deepstream/deepstream-5.1/lib

    

1.3 安装 DeepStream SDK

• Method 1: 使用 SDK Manager

  在安装 JetPack 4.5.1 时从 Additional SDKs 中选择  DeepStreamSDK。 

• Method 2: 使用 DeepStream tar 压缩包安装

  1. 下载 DeepStream 5.1 Jetson tar  压缩包 eepstream_sdk_v5.1.0_jetson.tbz2 到 Jetson 设备。

  2. 输入下面的命令解压并安装 DeepStream SDK：

     $ sudo tar -xvf deepstream_sdk_v5.1.0_jetson.tbz2 -C /
     $ cd /opt/nvidia/deepstream/deepstream-5.1
     $ sudo ./install.sh
     $ sudo ldconfig

      

亲测这个方法好使，后面的方法就不多介绍了。

• Method 3: Using the DeepStream Debian package

  Download the DeepStream 5.1 Jetson Debian package deepstream-5.1_5.1.0-1_arm64.deb, to the Jetson device. Then enter the command:

  $ sudo apt-get install ./deepstream-5.1_5.1.0-1_arm64.deb

  Note

  If you install the DeepStream SDK Debian package using the dpkg command, you must install the following packages before installing the debian package:

  • libgstrtspserver-1.0-0

  • libgstreamer-plugins-base1.0-dev

• Method 4: Using the apt-server

  1. Open the apt source configuration file in a text editor, using a command similar to $ sudo vi /etc/apt/sources.list.d/nvidia-l4t-apt-source.list

  2. Change the repository name and download URL in the deb commands shown below: deb https://repo.download.nvidia.com/jetson/common r32.5 main

  3. Save and close the source configuration file.

  4. Enter the commands:

     $ sudo apt update
     $ sudo apt install deepstream-5.1
     

• Method 5: Use Docker container DeepStream docker containers are available on NGC. See the Docker Containers section to learn about developing and deploying DeepStream using docker containers.

1.4 运行 deepstream-app (the reference application——参考应用)

1. 导航到开发套件的样例目录。

2. 输入下面的命令运行样例程序：

   $ deepstream-app -c <path_to_config_file>

   Where <path_to_config_file> is the pathname of one of the reference application’s configuration files, found in configs/deepstream-app/. See Package Contents for a list of the available files.

   Note

   • You can find sample configuration files under /opt/nvidia/deepstream/deepstream-5.1/samples directory. Enter this command to see application usage:

     $ deepstream-app --help
     

   • To save TensorRT Engine/Plan file, run the following command:

     $ sudo deepstream-app -c <path_to_config_file>
     

1. To show labels in 2D Tiled display view, expand the source of interest with mouse left-click on the source. To return to the tiled display, right-click anywhere in the window.

2. Keyboard selection of source is also supported. On the console where application is running, press the z key followed by the desired row index (0 to 9), then the column index (0 to 9) to expand the source. To restore 2D Tiled display view, press z again.

... ...

 

Jetson Nano 已经安装了 GStreamer，但是，nVidia 并没有完成配置， include 路径需要自己在配置一下才行。方法如下：

在 /etc/profile 文件最后添加以下两行内容即可：

export C_INCLUDE_PATH=/usr/include/gstreamer-1.0:/usr/include/glib-2.0:/usr/lib/aarch64-linux-gnu/glib-2.0/include:$C_INCLUDE_PATH
export CPLUS_INCLUDE_PATH=/usr/include/gstreamer-1.0:/usr/include/glib-2.0:/usr/lib/aarch64-linux-gnu/glib-2.0/include:$CPLUS_INCLUDE_PATH

保存后重启系统，即可。用下面的程序（文件名 basic-tutorial-1.c ）测试一下：

#include <gst/gst.h>

int
main (int argc, char *argv[])
{
  GstElement *pipeline;
  GstBus *bus;
  GstMessage *msg;

  /* Initialize GStreamer */
  gst_init (&argc, &argv);

  /* Build the pipeline */
  pipeline =
      gst_parse_launch
      ("playbin uri=https://www.freedesktop.org/software/gstreamer-sdk/data/media/sintel_trailer-480p.webm",
      NULL);

  /* Start playing */
  gst_element_set_state (pipeline, GST_STATE_PLAYING);

  /* Wait until error or EOS */
  bus = gst_element_get_bus (pipeline);
  msg =
      gst_bus_timed_pop_filtered (bus, GST_CLOCK_TIME_NONE,
      GST_MESSAGE_ERROR | GST_MESSAGE_EOS);

  /* Free resources */
  if (msg != NULL)
    gst_message_unref (msg);
  gst_object_unref (bus);
  gst_element_set_state (pipeline, GST_STATE_NULL);
  gst_object_unref (pipeline);
  return 0;
}

用下面的命令编译：

gcc basic-tutorial-1.c -o basic-tutorial-1 `pkg-config --cflags --libs gstreamer-1.0`

运行编译好的程序，应该能看到来自 https://www.freedesktop.org/software/gstreamer-sdk/data/media/sintel_trailer-480p.webm一段很有趣的视频。

其实到此为止，已经可以正常工作了，本文就先写这么多，谢谢阅读！