Installation Error: ‘H5Fis_hdf5’ was not declared in this scope

[sue k]

hello,  I tried to install Caffe on Ubuntu 18.04. However I am getting the following error: 

src/caffe/net.cpp: In member function ‘void caffe::Net<Dtype>::CopyTrainedLayersFrom(const string&)’:

src/caffe/net.cpp:774:7: error: there are no arguments to ‘H5Fis_hdf5’ that depend on a template parameter, so a declaration of ‘H5Fis_hdf5’ must be available [-fpermissive]

   if (H5Fis_hdf5(trained_filename.c_str())) {

       ^~~~~~~~~~

src/caffe/net.cpp:774:7: note: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)

src/caffe/net.cpp: In instantiation of ‘void caffe::Net<Dtype>::CopyTrainedLayersFrom(const string&) [with Dtype = float; std::__cxx11::string = std::__cxx11::basic_string<char>]’:

src/caffe/net.cpp:993:1:   required from here

src/caffe/net.cpp:774:17: error: ‘H5Fis_hdf5’ was not declared in this scope

   if (H5Fis_hdf5(trained_filename.c_str())) {

       ~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~

src/caffe/net.cpp: In instantiation of ‘void caffe::Net<Dtype>::CopyTrainedLayersFrom(const string&) [with Dtype = double; std::__cxx11::string = std::__cxx11::basic_string<char>]’:

src/caffe/net.cpp:993:1:   required from here

src/caffe/net.cpp:774:17: error: ‘H5Fis_hdf5’ was not declared in this scope

Makefile:576: recipe for target '.build_release/src/caffe/net.o' failed

make: *** [.build_release/src/caffe/net.o] Error 1

make: *** Waiting for unfinished jobs....

have you ever met this problem?

[Sss Kkk]

check this issue

https://github.com/BVLC/caffe/issues/6550

BTW, I have had much better luck running docker instances. NVidia has a GPU cloud website from which you can pull pre-configured containers with in built CUDA, Caffe, Tensorflow, jupyter, etc.

Also, there is an Nvidia-docker  application that allows docker containers to access the Nvidia gpu in your system. I found it to be as fast as running caffe on a bare metal install.

[MANIDEEP Allenki]

I am also facing same issue at the moment. I am using Ubuntu 19.04 and doing for CPU only. Please provide some solution if you found any

[Rusty Screws]

First step .... how many installation of python do you have installed? And which versions .....
Suggestion make sure you only have one version installed.  Make sure it is up to date and and pip is up to date ...
The easiest way to get things messed up ... like I have done is to not plan your installation. 
The key to a successful installation is to make sure all your directory paths are correct.
follow these instructions ...
https://garywoodfine.com/configure-python-3-4-default-ubuntu-14-04-2/