当前位置：网站首页>安装C3D v1.0

安装C3D v1.0

2022-07-20 21:38:00 【MallocLu】

前言

主要参考教程

可能用到的文件或压缩包：

链接：https://pan.baidu.com/s/1vIRxSJHRoBnrnsjczISOPg?pwd=7asr
提取码：7asr
复制这段内容后打开百度网盘手机App，操作更方便哦

下面所有命令均在root权限下执行，若提示权限不足，则在前面加sudo

主要参考教程若网址失效，可以在网盘里看到“主要参考教程网页备份”

本文所介绍的是C3D v1.0，若想使用C3D v1.1，则其和caffe安装的过程一模一样，跟着“个人caffe安装教程”走一遍，然后再安装C3D v1.1即可

安装CUDA&CUDNN

安装CUDA 10.2及对应的CUDNN（参考“主要参考教程”），为了简化操作作者使用了“矩池云平台”：

已经配置好了CUDA和cuDNN，就不用自己配置了。开一个1元/时的环境就够用

安装依赖

apt-get update
apt-get install -y libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler
apt-get install --no-install-recommends libboost-all-dev
apt-get install -y libopenblas-dev liblapack-dev libatlas-base-dev libgflags-dev libgoogle-glog-dev liblmdb-dev git cmake build-essential

# 打开环境配置文件
vim ~/.bashrc

# 在打开的文件末尾添加下面两行
export LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH

# 使环境配置文件生效
source ~/.bashrc

安装Opencv

将opencv-3.4.10.zip复制到/root目录下
unzip opencv-3.4.10.zip
删除opencv-3.4.10.zip（为了节省空间）

执行下面的命令

# 进入文件夹
cd opencv-3.4.10
# 创建build文件夹
mkdir build
# 进入build文件夹
cd build
# 编译
cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..
make
# 安装
make install
# 查看是否安装成功 成功将输出：3.4.10
pkg-config --modversion opencv

默认Python2

alias python=python2
# 输出python版本为2.?则为成功
python -V

安装boost

将boost_1_72_0.7z复制到/root目录下
apt-get install p7zip-full
7z x boost_1_72_0.7z
删除boost_1_72_0.7z（为了节省空间）

执行下面的命令

# 进入文件夹
cd boost_1_72_0.7z

./bootstrap.sh --with-libraries=all 
./b2
./b2 install

安装C3D

将C3D-master.zip复制到/root目录下
unzip C3D-master.zip
删除C3D-master.zip（为了节省空间）

执行下面的命令

# 进入文件夹
cd C3D-master/C3D-v1.0/python
# 安装依赖包
for req in $(cat requirements.txt); do pip install $req; done
# 返回上一级目录
cd ..
# 复制文件
cp Makefile.config.example Makefile.config

修改文件

1）修改Makefile.config

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 
修改为
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
		
# OPENCV_VERSION := 3
修改为
OPENCV_VERSION := 3

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib 
修改为
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 /usr/lib/x86_64-linux-gnu/hdf5/serial

2）修改Makefile

NVCCFLAGS +=-ccbin=$(CXX) -Xcompiler-fPIC $(COMMON_FLAGS)
修改为
NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)

LIBRARIES := cudart cublas curand \
	pthread \
	glog protobuf leveldb snappy \
	boost_system \
	hdf5_hl hdf5 \
	opencv_core opencv_highgui opencv_imgproc
修改为
LIBRARIES := cudart cublas curand \
	pthread \
	glog protobuf leveldb snappy \
	boost_system \
	hdf5_serial_hl hdf5_serial \
	opencv_core opencv_highgui opencv_imgproc

3）修改/src/caffe/test/test_gradient_check_util.hpp

Dtype scale = std::max(
std::max(fabs(computed_gradient), fabs(estimated_gradient)), 1.);
修改为
Dtype scale = std::max(
std::max(fabs(computed_gradient), fabs(estimated_gradient)),
Dtype(1.));

编译

make all
# 可能有错误：recipe for target 'build/examples/mnist/convert_mnist_data.o' failed
# 不用管

make test

make runtest
# 可能有错误，不用管
# Makefile:285: recipe for target 'runtest' failed
# make: *** [runtest] Error 127

C3D测试

将文件conv3d_deepnetA_sport1m_iter_1900000复制到C3D-master/C3D-v1.0/examples/c3d_feature_extraction/
sh c3d_sport1m_feature_extraction_frm.sh
出现错误：…/…/build/tools/extract_image_features.bin: error while loading shared libraries: libopencv_core.so.3.4: cannot open shared object file: No such file or directory
解决：在/etc/ld.so.conf.d/下创建OpenCV.conf，内容为/usr/local/lib，然后执行命令sudo ldconfig
sh c3d_sport1m_feature_extraction_frm.sh
执行成功，E0707 14:27:41.620281 36381 extract_image_features.cpp:121] Successfully extracted 16 features!