caffe(GPU)版编译安装

ubuntu22.04，python3.7，cuda11.1，cudnn8.1.1，opencv4.4.0

1.opencv4.4.0安装：

https://blog.csdn.net/qq_43552048/article/details/132069703

2.安装依赖

sudo apt-get install cmake git unzip -y
sudo apt-get install libprotobuf-dev libleveldb-dev liblmdb-dev -y
sudo apt-get install libsnappy-dev libhdf5-serial-dev protobuf-compiler -y
sudo apt-get install --no-install-recommends libboost-all-dev -y
sudo apt-get install libatlas-base-dev libopenblas-dev -y
sudo apt-get install the python3-dev python3-skimage -y
sudo pip3 install pydot -y
sudo apt-get install graphviz -y

3.下载caffe

wget -O caffe.zip https://github.com/Qengineering/caffe/archive/ssd.zip

注意这个是别人改版的，caffe官方github已经停止更新，这是官方的：https://github.com/BVLC/caffe

4.构建

解压后进入工程目录：

cd caffe-ssd

修改Makefile.config.example文件名为Makefille.config，还需修改相应的变量，这是我的仅供参考：

## Refer to http://caffe.berkeleyvision.org/installation.html
# Contributions simplifying and improving our build system are welcome!

# snapshot feature maps inputted to conv and fc layers
# USE_SNAPSHOT_FEATURE := 1

# display profiling results
# USE_PROFILE_DISPLAY := 1

# cuDNN acceleration switch (uncomment to build with cuDNN).
USE_CUDNN := 1
WITH_PYTHON_LAYER := 1

# CPU-only switch (uncomment to build without GPU support).
# CPU_ONLY := 1

# uncomment to disable IO dependencies and corresponding data layers
USE_OPENCV := 1
# USE_LEVELDB := 0
# USE_LMDB := 0

# uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)
#	You should not set this flag if you will be reading LMDBs with any
#	possibility of simultaneous read and write
# ALLOW_LMDB_NOLOCK := 1

# Uncomment if you're using OpenCV 3
OPENCV_VERSION := 4

# To customize your choice of compiler, uncomment and set the following.
# N.B. the default for Linux is g++ and the default for OSX is clang++
# CUSTOM_CXX := g++

# CUDA directory contains bin/ and lib/ directories that we need.
CUDA_DIR := /usr/local/cuda
# On Ubuntu 14.04, if cuda tools are installed via
# "sudo apt-get install nvidia-cuda-toolkit" then use this instead:
# CUDA_DIR := /usr

# CUDA architecture setting: going with all of them.
# For CUDA < 6.0, comment the *_50 lines for compatibility.
# CUDA_ARCH := #-gencode arch=compute_20,code=sm_20 \
		#-gencode arch=compute_20,code=sm_21 \
#		-gencode arch=compute_30,code=sm_30 \
#		-gencode arch=compute_35,code=sm_35 \
#		-gencode arch=compute_50,code=sm_50 \
#		-gencode arch=compute_50,code=compute_50
		
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_70,code=sm_70 \
             -gencode arch=compute_80,code=sm_80


# BLAS choice:
# atlas for ATLAS (default)
# mkl for MKL
# open for OpenBlas
BLAS := atlas
# Custom (MKL/ATLAS/OpenBLAS) include and lib directories.
# Leave commented to accept the defaults for your choice of BLAS
# (which should work)!
# BLAS_INCLUDE := /path/to/your/blas
# BLAS_LIB := /path/to/your/blas

# Homebrew puts openblas in a directory that is not on the standard search path
# BLAS_INCLUDE := $(shell brew --prefix openblas)/include
# BLAS_LIB := $(shell brew --prefix openblas)/lib

# This is required only if you will compile the matlab interface.
# MATLAB directory should contain the mex binary in /bin.
# MATLAB_DIR := /usr/local
# MATLAB_DIR := /Applications/MATLAB_R2012b.app

# NOTE: this is required only if you will compile the python interface.
# We need to be able to find Python.h and numpy/arrayobject.h.
PYTHON_INCLUDE := /usr/local/include/python3.7m \
		/usr/local/lib/python3.7/site-packages/numpy/core/include
# Anaconda Python distribution is quite popular. Include path:
# Verify anaconda location, sometimes it's in root.
# ANACONDA_HOME := $(HOME)/anaconda
# PYTHON_INCLUDE := $(ANACONDA_HOME)/include \
		# $(ANACONDA_HOME)/include/python2.7 \
		# $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \

# Uncomment to use Python 3 (default is Python 2)
PYTHON_LIBRARIES := boost_python python3.7m
# PYTHON_INCLUDE := /usr/include/python3.5m \
#                 /usr/lib/python3.5/dist-packages/numpy/core/include

# We need to be able to find libpythonX.X.so or .dylib.
PYTHON_LIB := /usr/local/lib
# PYTHON_LIB := $(ANACONDA_HOME)/lib

# Homebrew installs numpy in a non standard path (keg only)
# PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include
# PYTHON_LIB += $(shell brew --prefix numpy)/lib

# Uncomment to support layers written in Python (will link against Python libs)
# WITH_PYTHON_LAYER := 1

# Whatever else you find you need goes here.
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial 
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial 
# If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies
# INCLUDE_DIRS += $(shell brew --prefix)/include
# LIBRARY_DIRS += $(shell brew --prefix)/lib

# Uncomment to use `pkg-config` to specify OpenCV library paths.
# (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)
USE_PKG_CONFIG := 1

# N.B. both build and distribute dirs are cleared on `make clean`
BUILD_DIR := build
DISTRIBUTE_DIR := distribute

# Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171
# DEBUG := 1

# The ID of the GPU that 'make runtest' will use to run unit tests.
TEST_GPUID := 0

# enable pretty build (comment to see full commands)
Q ?= @

注意一定要使用非base环境，即退出anaconda，这里需要特别注意的是CUDA_ARCH，PYTHON_INCLUDE，PYTHON_LIBRARIES
我的是RTX30系列显卡，CUDA_ARCH注释了30和35那两行，不然后面编译会报错nvcc fatal : Unsupported gpu architecture 'compute_30'，PYTHON_INCLUDE和PYTHON_LIBRARIES改成你非anaconda里的python，PYTHON_INCLUDE需要填找到你python3版本对应安装的numpy。PYTHON_LIBRARIES那个boost_python3应该是说在你的usr/local/lib下有libboost_python3.so这个文件，这一般是个软连接文件链接到你安装的那个python版本的boost库文件，比如我的是python3.7就是链接到libboost_python37.so。上面安装依赖sudo apt install libboost-all-dev可能下载的是你系统里其他python3版本的boost.so库，你的boost如果没有对应上你的python3.x版本就需要去官网下载下来源码编译：

boost官方：https://www.boost.org/users/history/version_1_68_0.html
下载boost_1_68_0.tar.gz或可能其它版本

编译boost：

tar xzvf boost_1_68_0.tar.gz
cd boost_1_68_0/
sudo apt-get update  
sudo apt-get install build-essential autotools-dev libicu-dev ibbz2-dev
#卸载旧版本
# uninstall dpkg
sudo apt --purge remove libboost-dev
sudo apt --purge remove libboost-all-dev
sudo apt --purge autoremove libboost-all-dev
 
# to uninstall the version which we installed from source
sudo rm -rf /usr/lib/libboost_*
sudo rm -rf /usr/include/boost
 
./bootstrap.sh
./b2 --with-python include="/usr/local/include/python3.7m/" 
sudo ./b2 install -j4 
 
# Add the Boost libraries path to the default Ubuntu library search path
sudo /bin/bash -c 'echo "/usr/local/lib" > /etc/ld.so.conf.d/boost.conf'
 
sudo  ldconfig
 
#查看 boost 版本信息
cat /usr/local/include/boost/version.hpp | grep "BOOST_LIB_VERSION"
#输出如：
#  //  BOOST_LIB_VERSION must be defined to be the same as BOOST_VERSION
#  define BOOST_LIB_VERSION "1_68"
# 编译安装成功后，在 /usr/local/lib 中会生成有：libboost_python37.so 和 libboost_python37.a 等文件.
cd /usr/local/lib  
sudo ln -s libboost_python-py37.so libboost_python3.so  
sudo ln -s libboost_python-py37.a libboost_python3.a  
 
# 建立到 /usr/lib/x86_64-linux-gnu 路径的软连接
sudo cp /usr/local/lib/libboost_python37.a  /usr/lib/x86_64-linux-gnu/libboost_python-py37.a
sudo cp /usr/local/lib/libboost_python37.so.1.68.0  /usr/lib/x86_64-linux-gnu/libboost-py37.so
sudo ln -s /usr/lib/x86_64-linux-gnu/libboost-py37.so libboost_python.so

参考：https://blog.csdn.net/jiangbohan789/article/details/105115998/

之后开始构建：

make clean
make all -j$(nproc)

这里可能会报感应不到opencv的问题：
1.src/caffe/util/io.cpp:6:10: fatal error: opencv2/core/core.hpp: No such file or directory
在/usr/local/include文件夹发现是opencv4，在它子文件夹是opencv2，把opencv2复制出来和opencv4同一个文件夹就可以解决。

2.Package opencv was not found in the pkg-config search path.

cd /usr/local/lib
sudo mkdir pkgconfig
cd pkgconfig
sudo touch opencv.pc

在opencv.pc中添加以下信息，注意这些信息需要与自己安装opencv时的库路径对应：

prefix=/usr/local
exec_prefix=${prefix}
includedir=${prefix}/include
libdir=${exec_prefix}/lib

Name: opencv
Description: The opencv library
Version:4.0.1
Cflags: -I${includedir}/opencv4
Libs: -L${libdir} -lopencv_shape -lopencv_stitching -lopencv_objdetect -lopencv_superres -lopencv_videostab -lopencv_calib3d -lopencv_features2d -lopencv_highgui -lopencv_videoio -lopencv_imgcodecs -lopencv_video -lopencv_photo -lopencv_ml -lopencv_imgproc -lopencv_flann  -lopencv_core
~                                               

参考：
https://blog.csdn.net/l2181265/article/details/107080398/
https://blog.csdn.net/PecoHe/article/details/97476135

3./usr/bin/ld: cannot find -lboost_python3: No such file or directory
参考上面编译安装boost
4. /usr/bin/ld: /usr/lib/gcc/x86_64-linux-gnu/7/../../../x86_64-linux-gnu/libboost_python.so: undefined reference to PyCMethod_New
这个是因为我之前将libboost_python.so链接到python3.10版本的libboost导致的，必须链接到libboost-py37.so

构建顺利的话就显示：
若很多错就输入make all -j$(nproc) | grep error查错就行了
最后使用python构建：

make pytest

成功最后显示：

5.添加环境变量

cd ~
sudo gedit ~/.bashrc

添加：

export PYTHONPATH="${PYTHONPATH}:$HOME/caffe/caffe-ssd/python"

6.测试

这里我好像还少了skimage和protobuf等一些包，记得之前我是sudo apt install安装了依赖的，发现是因为这些包安装在/usr/lib/python3/dist-packages里而系统搜索路径好像并没有包含这个路经：
这里我至显示/usr/loca/…下的一些路径，解决办法：

which pip
which pip3

因此使用/usr/local/bin下的pip3 install XXX安装。

总体非常麻烦，整了可能快两天，若能帮到你的话麻烦点个赞求安慰一下… 最后感谢：
https://blog.csdn.net/weixin_39161727/article/details/120136500（cpu版）
https://qengineering.eu/install-caffe-on-ubuntu-20.04-with-opencv-4.4.html
https://www.bilibili.com/video/BV1oA411j7Cf/