TensorRT部署

下载

在 https://developer.nvidia.com/nvidia-tensorrt-7x-download 选择对应设备型号
LINUX查看CUDA版本 cat /usr/local/cuda/version.txt
LUNUX查看CUDNN版本 cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
下载对应Tar文件

安装

tar -xzvf TensorRT-XXX.tar.gz

然后设置路径

nano ~/.bashrc
# 加入以下
export LD_LIBRARY_PATH=TensorRT解压路径/lib:$LD_LIBRARY_PATH
# 然后生效
source ~./bashrc

安装Python接口

cd TensorRT-7.0.0.11/python
sudo pip3 install tensorrt-*-cp3x-none-linux_x86_64.whl
cd ../uff
sudo pip3 install uff-0.6.5-py2.py3-none-any.whl
cd ../graphsurgeon
sudo pip3 install graphsurgeon-0.4.1-py2.py3-none-any.whl

验证安装