coremltool converting Caffe model: "RuntimeError: Unable to load Caffe converter library."

Question

I'm trying to convert a caffemodel file with the coremltools, and get the following:

>>> import coremltools
>>> coreml_model = coremltools.converters.caffe.convert(('model.caffemodel','deploy.prototxt','mean.binaryproto'))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Library/Python/2.7/site-packages/coremltools/converters/caffe/_caffe_converter.py", line 130, in convert
    predicted_feature_name)
  File "/Library/Python/2.7/site-packages/coremltools/converters/caffe/_caffe_converter.py", line 144, in _export
    raise RuntimeError('Unable to load Caffe converter library.')
RuntimeError: Unable to load Caffe converter library.

I installed coremltools with pip:

sudo pip install --ignore-installed six coremltools

I was getting pip errors with the six package, hence the extra argument. The documentation doesn't say I need an additional packages for caffe. Even the example code in the documentation:

mport coremltools 
# Convert a caffe model to a classifier in Core ML
coreml_model = coremltools.converters.caffe.convert(('bvlc_alexnet.caffemodel', 'deploy.prototxt'), predicted_feature_name='class_labels.txt' )

failes to run with this error.

Core ML

3.8k

Posted by

roscott

Reply

Add a Comment

Answer 1

You are using the wrong version of Python (and possibly Keras).

It is best if you create a virtualenv and install the dependencies in there:

virtualenv -p /usr/bin/python2.7 env
source env/bin/activate
pip install tensorflow
pip install keras==1.2.2
pip install coremltools

Notice that I'm specifying "-p /usr/bin/python2.7" so that the virtualenv uses the system Python, not whatever other Python you have installed.

Posted by

kerfuffle

Add a Comment

Answer 2

My installation sudo -H pip install --no-cache-dir six==1.4.1

Posted by

cktung

Add a Comment

Answer 3

Part of the issue was my install command. Just reinstalled with:

pip install -U coremltools

as listed in the documentation. Now I get the following:

Python 2.7.13 (default, Apr  4 2017, 08:47:57)
[GCC 4.2.1 Compatible Apple LLVM 8.1.0 (clang-802.0.38)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import coremltools
>>> coreml_model = coremltools.coverters.caffe.convert(('bclc_alexnet.caffemodel', 'deploy.prototxt'), predicted_feature_name='class_labels.txt')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: 'module' object has no attribute 'coverters'
>>> coreml_model = coremltools.converters.caffe.convert(('bclc_alexnet.caffemodel', 'deploy.prototxt'), predicted_feature_name='class_labels.txt')
Fatal Python error: PyThreadState_Get: no current thread
[1]    97245 abort      python

So now it is crashing Python, but the original error is gone

Posted by

roscott

Add a Comment

Answer 4

You are using the wrong version of Python (and possibly Keras).

It is best if you create a virtualenv and install the dependencies in there:

virtualenv -p /usr/bin/python2.7 env
source env/bin/activate
pip install tensorflow
pip install keras==1.2.2
pip install coremltools

Notice that I'm specifying "-p /usr/bin/python2.7" so that the virtualenv uses the system Python, not whatever other Python you have installed.

Posted by

kerfuffle

Add a Comment

Answer 5

Thanks, this solved the issue!

Posted by

roscott

Add a Comment

coremltool converting Caffe model: "RuntimeError: Unable to load Caffe converter library."

Accepted Reply

Replies