Community
cancel
Showing results for 
Search instead for 
Did you mean: 
FDeSi
Beginner
398 Views

Compiling opencv dace detector model

Hi, I'm try to compile this model for the intel compute stick.

I've run the download_weights.py for download the caffe model and i've run

mvNCCompile train.prototxt -w res10_300x300_ssd_iter_140000_fp16.caffemodel -o face_graph

but i've this error:

/usr/lib/python3/dist-packages/scipy/stats/morestats.py:16: DeprecationWarning: Importing from numpy.testing.decorators is deprecated, import from numpy.testing instead. from numpy.testing.decorators import setastest /usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88 return f(*args, **kwds) mvNCCompile v02.00, Copyright @ Intel Corporation 2017   WARNING: Logging before InitGoogleLogging() is written to STDERR F0613 09:57:36.231927 192 insert_splits.cpp:29] Unknown bottom blob 'data' (layer 'data_bn', bottom index 0) *** Check failure stack trace: *** Aborted (core dumped)

I don't know how to resolve. Have somebody some clue?

0 Kudos
4 Replies
David_V_Intel
Employee
152 Views

Hello FDeSi,

 

Thank you for posting on the Intel ® communities.

 

Before I can assist you further I will need to know the exact model of Intel ® Compute Stick you have.

 

 

Regards,

David V

 

Intel Customer Support Technician

Under Contract to Intel Corporation

FDeSi
Beginner
152 Views

Hi,

Thank you for reply.

I've the Movidius Neural Compute Stick, the first model, but i'm using the ncsdk2 in a docker image.

David_V_Intel
Employee
152 Views

Hello FDeSi,

 

Thank you for your response.

 

I am changing the topic to the appropriate support forum, you will get assistance as soon as possible.

 

 

Regards,

David V

Intel Customer Support Technician

Under Contract to Intel Corporation

Sahira_Intel
Moderator
152 Views

Hi Francesco,

 

I apologize for the delay in my response. This error is caused by compiling a deploy-mode model with a train-mode weight. Instead of using train.prototxt, use deploy.prototxt.

 

Best Regards,

Sahira

Reply