Jetson Xavier NX配置YOLOV5环境

一、pip3及依赖库安装

sudo apt-get update
sudo apt-get install python3-pip python3-dev -y
sudo apt-get install build-essential make cmake cmake-curses-gui -y
sudo apt-get install git g++ pkg-config curl -y
sudo apt-get install libatlas-base-dev gfortran libcanberra-gtk-module libcanberra-gtk3-module -y
sudo apt-get install libhdf5-serial-dev hdf5-tools -y
sudo apt-get install nano locate screen -y
sudo apt-get install libfreetype6-dev -y
sudo apt-get install protobuf-compiler libprotobuf-dev openssl -y
sudo apt-get install libssl-dev libcurl4-openssl-dev -y
sudo apt-get install cython3 -y

为了方便查看相关系统信息：

sudo -H pip3 install jetson-stats

安装完成后，重启系统，可以终端输入jtop调用

二、Cmake

wget http://www.cmake.org/files/v3.13/cmake-3.13.0.tar.gz
tar xpvf cmake-3.13.0.tar.gz cmake-3.13.0/
cd cmake-3.13.0/
./bootstrap --system-curl	
make -j4 
echo 'export PATH=~/cmake-3.13.0/bin/:$PATH' >> ~/.bashrc
source ~/.bashrc

Ｐｓ：如果出现

缺少：CURL_LIBRARY CURL_INCLUDE_DIR

sudo apt-get install curl
sudo apt-get install libssl-dev libcurl4-openssl-dev

大容量设备无法挂载

sudo apt-get install exfat-utils

完成以后，可以试一下：

cmake --version

可以看到安装完成的cmake版本信息

三、CUDA10.2、CUDNN8.0安装

2.1 CUDA安装

将Cuda10.2文件移到/usr/local中

sudo cp -r [目标文件的位置] [目标位置]

例如，我的文件原路径是：Download/cuda-10.2, 我的指令就应该是sudo cp -r Download/cuda-10.2 /usr/local
然后将/usr/local中的cuda-10.2中的targets解压，

sudo tar xzvf targets.tar

创建软连接：

sudo ln -s /usr/local/cuda-10.2/targets/aarch64-linux/lib lib64

sudo ln -s /usr/local/cuda-10.2/targets/aarch64-linux/include include

sudo chmod -R +x /usr/local/cuda-10.2

sudo ln -s /usr/local/cuda-10.2 /usr/local/cuda

sudo ldconfig

环境变量配置：

sudo vim ~/.bashrc

键盘输入a将进入Insert模式，然后将下方信息复制到文件末尾

export CUBA_HOME=/usr/local/cuda-10.2
export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64:$LD_LIBRARY_PATH
export PATH=/usr/local/cuda-10.2/bin:$PATH

然后esc退出Insert模式，然后键盘输入:wq 保存并退出
更新一下：

source ~/.bashrc

再安装一些依赖：

sudo apt-get update
sudo apt-get install cuda-toolkit-10-2

然后可以验证一下cuda：

nvcc -V

2.2 CUDNN安装

将cudnn文件中的三个deb文件拷到开发板里

然后依次执行：

sudo dpkg -i libcudnn8_8.0.0.180-1+cuda10.2_arm64.deb
sudo dpkg -i libcudnn8-dev_8.0.0.180-1+cuda10.2_arm64.deb
sudo dpkg -i libcudnn8-doc_8.0.0.180-1+cuda10.2_arm64.deb

然后修改安装路径：

sudo cp /usr/include/cudnn.h /usr/local/cuda/include/

sudo cp /usr/lib/aarch64-linux-gnu/libcudnn* /usr/local/cuda/lib64/

然后可以打开jtop看到

四、opencv

4.1 安装opencv的系统依赖和编解码库

sudo apt-get install build-essential -y
sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev -y
sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff5-dev libdc1394-22-dev -y
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev liblapacke-dev -y
sudo apt-get install libxvidcore-dev libx264-dev -y
sudo apt-get install libatlas-base-dev gfortran -y
sudo apt-get install ffmpeg -y

依赖安装中出现：

解决：

sudo ln -sf /usr/local/cuda/lib64/libcudnn_adv_infer.so.8.2.1 /usr/local/cuda/lib64/libcudnn_adv_infer.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn_ops_train.so.8.2.1 /usr/local/cuda/lib64/libcudnn_ops_train.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn.so.8.2.1 /usr/local/cuda/lib64/libcudnn.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn_ops_infer.so.8.2.1 /usr/local/cuda/lib64/libcudnn_ops_infer.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn_adv_train.so.8.2.1 /usr/local/cuda/lib64/libcudnn_adv_train.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn_cnn_infer.so.8.2.1 /usr/local/cuda/lib64/libcudnn_cnn_infer.so.8
 
sudo ln -sf /usr/local/cuda/lib64/libcudnn_cnn_train.so.8.2.1 /usr/local/cuda/lib64/libcudnn_cnn_train.so.8

4.2opencv安装

4.2.1 换源

sudo vim /etc/apt/sources.list

deb http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial main multiverse restricted universe
deb http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-backports main multiverse restricted universe
deb http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-proposed main multiverse restricted universe
deb http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-security main multiverse restricted universe
deb http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-updates main multiverse restricted universe
deb-src http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial main multiverse restricted universe
deb-src http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-backports main multiverse restricted universe
deb-src http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-proposed main multiverse restricted universe
deb-src http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-security main multiverse restricted universe
deb-src http://mirrors.ustc.edu.cn/ubuntu-ports/ xenial-updates main multiverse restricted universe

安装依赖

sudo apt-get update
sudo apt-get install libjasper1 libjasper-dev

4.2.2 安装

下载文件 opencv 与opencv_contrib 的源码，版本要一致

在openccv-4.1.1中新建build文件

mkdir build 
cd build 
cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local -D OPENCV_EXTRA_MODULES_PATH=/home/******/*****/opencv_contrib-4.1.1/modules ..

sudo make -j4

In file included from /home/htkt/7111Yolo/opencv-3.4.5/modules/stitching/include/opencv2/stitching.hpp:49:0,
from /home/htkt/7111Yolo/opencv-3.4.5/modules/stitching/src/precomp.hpp:59,
from /home/htkt/7111Yolo/opencv-3.4.5/build/modules/stitching/opencv_stitching_pch_dephelp.cxx:1:
/home/htkt/7111Yolo/opencv-3.4.5/modules/stitching/include/opencv2/stitching/detail/matchers.hpp:52:12: fatal error: opencv2/xfeatures2d/cuda.hpp: No such file or directory

include “opencv2/xfeatures2d/cuda.hpp”

        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ compilation terminated.

上面的ERROR，原因是include的cuda.hpp头文件路径找不到，需要改为绝对路径。打开它提示的文件matchers.hpp，然后找到你的cuda.hpp目录，如下所示。

复制你的路径，然后替换掉include “opencv2/xfeatures2d/cuda.hpp”，比如我的路径：/home/dzh/Software/opencv-3.4.5/opencv_contrib-3.4.5/modules/xfeatures2d/include/opencv2/xfeatures2d/cuda.hpp

fatal error: boostdesc_bgm.i: No such file or directory

https://blog.csdn.net/ben_xiao_hai_123/article/details/126262891

sudo make install

配置环境：

sudo gedit /etc/ld.so.conf.d/opencv.conf

在打开的文本中输入：

/usr/local/lib

然后保存、关闭

sudo ldconfig

五、Pytorch

将以下文件拷贝到开发板

依赖安装：

sudo apt-get install libopenmpi2
sudo apt-get install libopenblas-dev
sudo apt-get install libjpeg-dev zlib1g-dev
sudo pip3 install -i https://pypi.tuna.tsinghua.edu.cn/simple pillow

然后终端输入：

sudo pip3 install torch-1.8.0-cp36-cp36m-linux_aarch64.whl

六、torchvision

环境配置：

gedit ~/.bashrc

末尾添加：

export OPENBLAS_CORETYPE=ARMV8

source ~/.bashrc

cd到torchvision文件夹中，然后：

sudo python3 setup.py install

export BUILD_VERSION=0.9.0
python3 setup.py install --user

测试：

python3
import torch
import torchvision                          

import未报错就说明ok了

七、yolov5依赖安装

sudo pip3 install matplotlib==3.2.2 -i https://pypi.tuna.tsinghua.edu.cn/simple
sudo pip3 install --upgrade Cython	                            
sudo apt-get remove python-numpy
sudo pip3 install numpy==1.19.4
sudo pip3 install scipy==1.4.1.	     
sudo pip3 install tqdm==4.61.2
sudo pip3 install seaborn==0.11.1
sudo pip3 install scikit-build==0.11.1

sudo pip3 install tensorboard==2.5.0 -i https://pypi.tuna.tsinghua.edu.cn/simple

这里会因为python版本低于3.7报错，可以尝试降低protobuf的版本：

sudo pip3 install protobuf==3.19.6

sudo pip3 install PyYAML==5.4.1
sudo pip3 install thop
sudo pip3 install pycocotools

如果网不好会经常因为ReadTimeoutError下载失败，可以考虑：

sudo pip3 --default-timeout=1688 install matplotlib==3.2.2 -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn

–default-timeout=1688 就是让其检测延迟的时间变长，以防止因为网络问题而直接报错