安装必要的基础环境

安装Anaconda并添加环境变量：

sudo vim ~/.bashrc
export PATH=~/anaconda3/bin:$PATH

创建虚拟环境：conda create -n 名字 python=需要的版本
更新pip: python -m pip install --upgrade pip
更换conda和pip为国内源。
我的系统为Ubuntu 18.04，显卡为Nvidai Tesla T4（根据自己的显卡型号去灵活选择相关的驱动和CUDA等版本）

需要安装的运行环境

Ubuntu 18.04
GStreamer 1.14.1
NVIDIA Driver 460.32
CUDA 11.1
cuDNN 8.0.5
cuBLAS
TensorRT 7.2.3

1.清除原安装环境：

sudo rm -rf /usr/local/deepstream /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstnv* /usr/bin/deepstream* /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libnvdsgst*
/usr/lib/x86_64-linux-gnu/gstreamer-1.0/deepstream* /opt/nvidia/deepstream/deepstream*
sudo rm -rf /usr/lib/x86_64-linux-gnu/libv41/plugins/libcuvidv4l2_plugin.so

如果安装Deepstream之前版本，则先卸载

To remove DeepStream 4.0 or later installations:
Open the uninstall.sh file in /opt/nvidia/deepstream/deepstream/
Set PREV_DS_VER as 4.0
Run the following script as sudo: ./uninstall.sh

2.在安装DeepStream SDK之前，请输入以下命令安装必要的软件包

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

3.下载和安装NVIDIA驱动460.32 驱动下载

chmod 755 NVIDIA-Linux-x86_64-460.32.03.run
./NVIDIA-Linux-x86_64-460.32.03.run
nvidia-smi

4.下载和安装CUDA 11.111.5之前版本

wget https://developer.download.nvidia.com/compute/cuda/11.1.0/local_installers/cuda_11.1.0_455.23.05_linux.run
sudo sh cuda_11.1.0_455.23.05_linux.run
sudo vim ~/.bashrc  #末尾加入路径
export PATH=/usr/local/cuda-11.1/bin${PATH:+:${PATH}}  
export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}} 
source ~/.bashrc
nvcc -V #若安装正确则会显示CUDA 11.1

注：若是出现安装问题则可以尝试命令

sudo ./cuda_11.1.0_455.23.05_linux.run --librartpath=/usr/local/cuda-11.1

5.安装cudnn选择CUDA11.1对应的cudnn版本

tar -xzvf cudnn-11.2-linux-x86-v8.1.1.33.tgz(8.1兼容11.0、11.1、11.2)
sudo cp cuda/include/* /usr/local/cuda/include/
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64/
sudo chmod a+r /usr/local/cuda/include/cudnn.h
sudo chmod a+r /usr/local/cuda/lib64/libcudnn*
cat /usr/local/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2 # 验证
# 下载对应的libcudnn8_8.1.1.33-1+cuda11.2_amd64.deb   libcudnn8-dev_8.1.1.33-1+cuda11.2_amd64   libcudnn8-samples_8.1.1.33-1+cuda11.2_amd64.deb
sudo dpkg -i libcudnn8_8.1.1.33-1+cuda11.2_amd64.deb
sudo dpkg -i libcudnn8-dev_8.1.1.33-1+cuda11.2_amd64
sudo dpkg -i libcudnn8-samples_8.1.1.33-1+cuda11.2_amd64.deb
cd /usr/src/cudnn_samples_v8/mnistCUDNN
sudo make clean
sudo make #若出现缺少FreeImage.h报错，则执行 sudo apt-get install libfreeimage libfreeimage-dev 然后重新编译
./mnistCUDNN # 运行示例验证

CUDA和CUDNN安装成功

6.tensorRT安装选择TensorRT对应版本

tar -xzvf TensorRT-7.2.3.4.Ubuntu-18.04.x86_64-gnu.cuda-11.1.cudnn8.1.tar.gz #解压后复制到/home/用户名 目录下
cd ~/TensorRT-7.2.3.4/
sudo cp -r ./lib/* /usr/lib # 在TensorRT-7.2.3.4路径下执行
sudo cp -r ./include/* /usr/include # 在TensorRT-7.2.3.4路径下执行
sudo vim ~/.bashrc
#tensorrt7.2.3 need
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/用户名/TensorRT-7.2.3.4/lib #在最后添加路径
source ~/.bashrc
conda create -n 虚拟环境名字 python=3.8 #创建RT环境，安装可能需要的轮子包
conda activate 虚拟环境名字
cd ~/TensorRT-7.2.3.4/python
pip3 install tensorrt-7.2.3.4-cp38-none-linux_x86_64.whl
cd ~/TensorRT-7.2.3.4/graphsurgeon
pip3 install graphsurgeon-0.4.5-py2.py3-none-any.whl
cd ~/TensorRT-7.2.3.4/onnx_graphsurgeon
pip3 install onnx_graphsurgeon-0.2.6-py2.py3-none-any.whl
pip3 install pycuda
python3
import tensorrt
print（tensorrt.__version__）（7.2.3.4）

TensorRT安装成功

7.安装librdkafka

git clone https://github.com/edenhill/librdkafka.git
cd librdkafka
git reset --hard 7101c2310341ab3f4675fc565f64f0967e135a6a
./configure
make
sudo make install
sudo mkdir -p /opt/nvidia/deepstream/deepstream-5.1/lib
sudo cp /usr/local/lib/librdkafka* /opt/nvidia/deepstream/deepstream-5.1/lib

8.安装Deep Stream SDK下载文件

sudo tar -xvf deepstream_sdk_v5.1.0_x86_64.tbz2 -C /
cd /opt/nvidia/deepstream/deepstream-5.1/
sudo ./install.sh
sudo ldconfig
deepstream-app --version-all #若出现Gstreamer警告找不到libtritonserver.so文件：rm -rf ~/.cache/gstreamer-1.0/gst-inspect-1.0

Deepstream安装完成

9.验证Deepstream环境（测试官方Demo）

cd /opt/nvidia/deepstream/deepstream-5.1/samples/configs/deepstream-app
sudo vim source30_1080p_dec_infer-resnet_tracker_sgie_titled_display_int8.txt
#将sink0中的enable=1改为0；将sink1中的enable=0改为1，保存退出
#我这里遇到显示问题，不能够界面显示，所以用保存mp4的方法验证（没找到无法显示的解决办法...）
sudo deepstream-app -c source30_1080p_dec_infer-resnet_tracker_sgie_titled_display_int8.txt  #输出视频保存在同级目录下out.mp4