Caffe Installation Record

System: ubuntu16.04

Graphics card: GTX1060

cuda8.0,cudnn8.0, opencv3.1

Before the caffe,linux in Windows have tried, but did not succeed, so take advantage of this period of time to tackle the key.

After several toss, finally succeeded in setting up a good caffe in Ubuntu, this record the hole encountered, for inspection.

Please post the reference post, thank the experience of the great God

http://blog.csdn.net/shiorioxy/article/details/52652831

The first Direct installation Bo Master program installation, in make phase, appear

/usr/bin/ld:cannot Find-lopencv_imgcodecs

Collect2:error:ld returned 1 exit status

Makefile:572:recipe for target '. Build_release/lib/libcaffe.so.1.0.0-rc3 ' failed

After the search, it was found that OpenCV did not install the correct installation, due to the Linux system operation is not familiar. Helpless Reinstall the OPENCV

Reference: http://blog.csdn.net/xuzhongxiong/article/details/52717285

This time make passed, but there are new problems

In the process of compiling classification.o, there appears undefined reference to ' Cv::mat ... A crowd of similar error

It seems that many people have encountered this problem, so found a solution on the Internet

Added two lines to the Classification.cpp

#include <opencv2/imgproc/types_c.h>

#include <opencv2/objdetect/objdetect_c.h>

Originally thought to be all right here, who knows in the last make runtest, again out of the moth

The problem is described as follows:

. build_release/tools/caffe:error while loading shared libraries:libopencv_core.so.3.1:cannot open Shared object file:n o such file or directory.

Makefile:532:recipe for Target ' runtest ' failed.

Here I have been unable to spit, who told me that I did not study hard, the system reload over and over again, to this step is really do not want to do naught.

Thanks to the almighty Internet, this problem should not be the problem of Caffe itself, but Linux

Reference: http://blog.csdn.net/sahusoft/article/details/7388617

First to perform locate libopencv_core.so.3.1 and then specifically how to operate the Bo main article wrote very clearly.

Finally see Runtest success, I am almost excited to cry out of the feeling.

Although it may be a small step for Daniel, it is a 0 to 1 leap for me personally.