Original: https://blog.csdn.net/zong596568821xp/article/details/86077553
1. Install
#在home下新建文件夹，命名为tensorrt_tar，然后将下载的压缩文件拷贝进来解压
tar xzvf TensorRT-5.0.2.6.Ubuntu-16.04.4.x86_64-gnu.cuda-9.0.cudnn7.3.tar
#解压得到TensorRT-5.0.2.6的文件夹，将里边的lib绝对路径添加到环境变量中
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/lthpc/tensorrt_tar/TensorRT-5.0.2.6/lib
#安装TensorRT
cd TensorRT-5.0.2.6/python
sudo pip install tensorrt-5.0.2.6-py2.py3-none-any.whl
#安装UFF
cd TensorRT-5.0.2.6/uff
sudo pip install uff-0.5.5-py2.py3-none-any.whl
#安装graphsurgeon
cd TensorRT-5.0.2.6/graphsurgeon
sudo pip install graphsurgeon-0.3.2-py2.py3-none-any.whl
Conda can be installed in the environment, to facilitate isolation.
