WSL2 配置yolov5_caffe

一、安装配置Ubuntu

1、查看系统版本 

lsb_release -a            # 查看当前系统版本

2、换源

注意：需要根据步骤1查出的代号对下方源中的代号进行替换 

sudo vim /etc/apt/sources.list

# 阿里云镜像源：
deb http://mirrors.aliyun.com/ubuntu/ bionic main restricted
deb http://mirrors.aliyun.com/ubuntu/ bionic-updates main restricted
deb http://mirrors.aliyun.com/ubuntu/ bionic universe
deb http://mirrors.aliyun.com/ubuntu/ bionic-updates universe
deb http://mirrors.aliyun.com/ubuntu/ bionic multiverse
deb http://mirrors.aliyun.com/ubuntu/ bionic-updates multiverse
deb http://mirrors.aliyun.com/ubuntu/ bionic-backports main restricted universe multiverse
deb http://mirrors.aliyun.com/ubuntu/ bionic-security main restricted
deb http://mirrors.aliyun.com/ubuntu/ bionic-security universe
deb http://mirrors.aliyun.com/ubuntu/ bionic-security multiverse

sudo apt update
sudo apt upgrade 

3、安装基础工具

sudo apt install g++ git unzip pkg-config make cmake

4、安装samba服务

（1）安装samba

sudo apt install samba

（2）打开smb.conf文件

sudo vim /etc/samba/smb.conf

在文件最后加入以下配置并保存，然后退出

[work]  
    comment = samba home directory
    path = /home/lcat
    public = yes
    browseable = yes
    public = yes
    writeable = yes
    read only = no
    valid users = lcat
    create mask = 0777
    directory mask = 0777
    #force user = nobody
    #force group = nogroup
    available = yes

（3）设置新的samba密码

sudo smbpasswd -a lcat

（4）重新启动samba服务

sudo service smbd restart

（5）查看Ubuntu的IP地址

ifconfig -a

（6）window访问samba共享文件（映射网络驱动器）

二、配置环境

（一）安装CUDA

1、下载

 https://developer.nvidia.com/cuda-downloads 

2、安装

执行以下命令，除了是否安装驱动选择no外，其余均选择yes，默认安装路径

sudo sh cuda_10.0.130_410.48_linux.run
sudo sh cuda_10.0.130.1_linux.run

3、配置环境变量

vim ~/.bashrc
export PATH=/usr/local/cuda-10.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
source ~/.bashrc

4、验证安装

nvcc -V

（二）安装cuDNN

1、下载

 https://developer.nvidia.com/rdp/cudnn-archive 

2、解压并进入该文件夹下，执行如下命令，进行相关文件的复制：

tar -xzvf cudnn-10.0-linux-x64-v7.6.5.32.tgz

sudo cp cuda/include/cudnn.h /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*

3、打开终端，输入以下命令，查看cuDNN是否安装成功：

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

（三）安装anaconda

1、下载

wget https://repo.anaconda.com/archive/Anaconda3-5.2.0-Linux-x86_64.sh

2、安装

bash Anaconda3-5.2.0-Linux-x86_64.sh

一直Enter和yes，碰到安装Microsoft的VS code时选择No，结束安装

3、验证安装

conda -V

（四）安装opencv

1、下载

 https://opencv.org/releases/ 

2、安装

unzip opencv-3.4.2.zip
cd opencv-3.4.2
sudo mkdir build
cd build
sudo cmake ../
sudo make -j8
sudo make install

3、验证安装

pkg-config --modversion opencv     # 查看版本号
pkg-config --libs opencv           # 查看libs库

三、安装依赖

sudo apt install protobuf-compiler libhdf5-serial-dev 
sudo apt install libboost-all-dev libgflags-dev libgoogle-glog-dev libprotobuf-dev 
sudo apt install libatlas-base-dev libleveldb-dev liblmdb-dev libsnappy-dev

四、编译caffe

1、拉取yolov5_caffe源码

git clone https://github.com/Wulingtian/yolov5_caffe.git

2、进入yolov5_caffe目录，配置环境变量

cd yolov5_caffe
export CPLUS_INCLUDE_PATH=/home/lcat/anaconda3/include/python3.6m

3、编译

make all -j8

4、测试yolov5推理

修改 yolov5_caffe/tools/caffe_yolov5s.cpp 参数

cd tools
vim caffe_yolov5s.cpp

INPUT_W                模型输入宽度
INPUT_H                模型输入高度
NUM_CLASS              模型有多少个类别
NMS_THRESH             做非极大值抑制的阈值
CONF_THRESH            类别置信度
prototxt_path          caffe模型的prototxt路径
caffemodel_path        caffe模型的caffemodel路径
pic_path               测试图片路径

修改完成返回yolov5_caffe目录，编译执行

make -j8
cd build
./tools/caffe_yolov5s

运行成功将输出平均推理时间，并保存预测图片到当前目录下

五、错误

错误1

make: protoc: Command not found

解决：

sudo apt install protobuf-compiler

错误2

src/caffe/layers/hdf5_data_layer.cu:10:10: fatal error: hdf5.h: No such file or directory
 #include "hdf5.h"

解决：

sudo apt install libhdf5-serial-dev

错误3

./include/caffe/common.hpp:4:10: fatal error: boost/shared_ptr.hpp: No such file or directory
 #include <boost/shared_ptr.hpp>

解决：

sudo apt install libboost-all-dev

错误4

./include/caffe/common.hpp:5:10: fatal error: gflags/gflags.h: No such file or directory
 #include <gflags/gflags.h>

解决：

sudo apt install libgflags-dev

错误5

./include/caffe/common.hpp:6:10: fatal error: glog/logging.h: No such file or directory
 #include <glog/logging.h>

解决：

sudo apt install libgoogle-glog-dev

错误6

.build_release/src/caffe/proto/caffe.pb.h:9:10: fatal error: google/protobuf/stubs/common.h: No such file or directory
 #include <google/protobuf/stubs/common.h>

解决：

sudo apt install libprotobuf-dev

错误7

./include/caffe/util/mkl_alternate.hpp:14:10: fatal error: cblas.h: No such file or directory
 #include <cblas.h>

解决：

sudo apt install libatlas-base-dev

错误8

examples/siamese/convert_mnist_siamese_data.cpp:20:10: fatal error: leveldb/db.h: No such file or directory
 #include "leveldb/db.h"

解决：

sudo apt install libleveldb-dev

错误9

examples/mnist/convert_mnist_data.cpp:16:10: fatal error: lmdb.h: No such file or directory
 #include <lmdb.h>

解决：

sudo apt install liblmdb-dev

错误10

/usr/bin/ld: cannot find -lsnappy

解决：

sudo apt install libsnappy-dev

错误11

F0105 14:44:33.180271 26467 pooling_layer.cu:212] Check failed: error == cudaSuccess (48 vs. 0)  no kernel image is available for execution on the device

解决：根据安装的CUDA版本将 yolov5_caffe 中 Makefile.config 文件以下内容根据提示作出修改。本文将下列行：

CUDA_ARCH := -gencode arch=compute_61,code=compute_61

修改为：

CUDA_ARCH := -gencode arch=compute_30,code=sm_30 \
		-gencode arch=compute_35,code=sm_35 \
		-gencode arch=compute_50,code=sm_50 \
		-gencode arch=compute_52,code=sm_52 \
		-gencode arch=compute_60,code=sm_60 \
		-gencode arch=compute_61,code=sm_61 \
		-gencode arch=compute_61,code=compute_61