flownet2.0 caffe anaconda2 编译安装

---

1. 下载flownet2.0源码到指定目录

cd /home/zzq/saliency_models/deep_optical_flow
git clone https://github.com/lmb-freiburg/flownet2
cd flownet2
cp Makefile.config.example Makefile.config

---

2. 新建anaconda2 flownet2.0环境

cd ~/anaconda2/bin
conda create -n flownet2.0 python=2.7
source activate flownet2.0
conda install numpy
conda instaLL cython
conda install scipy
conda install scikit-image
pip install msgpack
pip install opencv-python

---

3. 配置flownet2.0环境

修改Makefile.config：

USE_CUDNN := 1
USE_OPENCV := 1
ANACONDA_HOME := $(HOME)/anaconda2/envs/flownet2.0
PYTHON_INCLUDE := $(ANACONDA_HOME)/include \
 		    $(ANACONDA_HOME)/include/python2.7 \
        	    $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \
                    /usr/include/python2.7 

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

修改Makefile：

LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial  opencv_core opencv_highgui opencv_imgproc opencv_imgcodecs

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

---

4. 编译flownet2.0

make -j `nproc` all tools pycaffe

如果出现fatal error: pyconfig.h: No such file or directory #include "pyconfig.h"，则：

make clean
export CPLUS_INCLUDE_PATH=/home/zzq/anaconda2/envs/flownet2.0/include/python2.7
make -j `nproc` all tools pycaffe

如果出现 ‘numeric_limits’ is not a member of ‘std’, 则在出错CPP文件中添加：

#include<stdexcept>
#include<limits>

如果出现

/usr/local/lib/libopencv_imgcodecs.so：对‘TIFFReadDirectory@LIBTIFF_4.0’未定义的引用
/usr/local/lib/libopencv_imgcodecs.so：对‘TIFFWriteEncodedStrip@LIBTIFF_4.0’未定义的引用
/usr/local/lib/libopencv_imgcodecs.so：对‘TIFFIsTiled@LIBTIFF_4.0’未定义的引用
/usr/local/lib/libopencv_imgcodecs.so：对‘TIFFOpen@LIBTIFF_4.0’未定义的引用

则在cmake opencv时添加：

-D BUILD_TIFF=ON

配置环境, 每次使用 flownet2.0 时, 都要进行如下操作

source set-env.sh

---

5. 测试flownet2.0

cd models,下载模型。

run-flownet.py ../models/FlowNet2/FlowNet2_weights.caffemodel.h5  ../models/FlowNet2/FlowNet2_deploy.prototxt.template /home/zzq/ZZQ/MSVSD/OUTPUT/video/1/1.jpg /home/zzq/ZZQ/MSVSD/OUTPUT/video/1/2.jpg /home/zzq/ZZQ/MSVSD/OUTPUT/video/1/flow2.flo 

可视化结果：

注意： conda install protobuf一定要在编译环境成功之后。否则会出现/usr/include/c++/5/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support must be enabled with the -std=c++11 or -std=gnu++11 compiler options. 的错误。