-
-
Notifications
You must be signed in to change notification settings - Fork 17
Undefined symbol in convert_imageset #33
Comments
Could you please include the result of |
The output of
|
Note that this recipe is only built for python 2.7, so you're actually using caffe from the default repositories. |
So the version from this feedstock is only for Python 2.7? For
I get And similarly Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/work/miniconda/envs/caffe_test/lib/python2.7/site-packages/caffe/__init__.py", line 1, in <module>
from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, AdaDeltaSolver, AdamSolver, NCCL, Timer
File "/work/miniconda/envs/caffe_test/lib/python2.7/site-packages/caffe/pycaffe.py", line 13, in <module>
from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, \
ImportError: libprotobuf.so.14: cannot open shared object file: No such file or directory The fix from #31 The output of
|
Those are because this recipe doesn't pin protobuf (a bug). Try |
Unfortunately, that did not help either. |
Oh, wait, it's looking for |
Alright so after doing
Only after |
That problem is because you have numpy installed from |
But how can I avoid this? Even if I set |
Please try |
The output of
|
Can you please share the most recent result of |
and
|
Sorry remind me what the error message is that you are getting now? Still this one or something else? |
Exactly, it's
but only after doing |
What if you just run |
That works without error... |
What about |
The result from |
We are rebuilding with the latest |
Sure, when will it be out? |
Should be out now. |
Seems to work fine for environments with python 2.7! Was this intended to work with python 3 as well, because this still gives errors |
No, conda-forge still doesn't build caffe for python 3, so that's a problem with the caffe from |
Alright, then I would consider this issue fixed. Thanks very much for your help and keep up the good work! |
Thanks for testing. :) |
I have a problem with the command line tools provided by caffe on Ubuntu 16.04
Doing
I get
convert_imageset: symbol lookup error: convert_imageset: undefined symbol: _ZN6google14FlagRegistererC1EPKcS2_S2_S2_PvS3_
The output of
$ c++filt _ZN6google14FlagRegistererC1EPKcS2_S2_S2_PvS3_
isgoogle::FlagRegisterer::FlagRegisterer(char const*, char const*, char const*, char const*, void*, void*)
The respective binary of
convert_imageset
is for me in$CONDA_ROOT/envs/test/bin
Since many tutorials or training instructions for caffe nets rely on these tools it would be great if they would work out of the box with the conda install.
The text was updated successfully, but these errors were encountered: