Community
cancel
Showing results for 
Search instead for 
Did you mean: 
park__kyungja
Beginner
224 Views

OpenVINO Model Optimizer Error of tensorflow model

Hi,

I'm trying to convert custom tensorflow  model to IR with Model Optimizer.

But, I've got error!

here are my command and error message.

/////////////////////////////////

C:\Program Files (x86)\IntelSWTools\openvino_2019.1.133\deployment_tools\model_optimizer>python mo_tf.py --input_model d:\openVINO\frozen_model_tf_1.14.pb --input keep_probabilty,input_image --input_shape (1),(1,40,40,1) --output inference/prediction --output_dir d:\openVINO
Model Optimizer arguments:
Common parameters:
        - Path to the Input Model:      d:\openVINO\frozen_model_tf_1.14.pb
        - Path for generated IR:        d:\openVINO
        - IR output name:       frozen_model_tf_1.14
        - Log level:    ERROR
        - Batch:        Not specified, inherited from the model
        - Input layers:         keep_probabilty,input_image
        - Output layers:        inference/prediction
        - Input shapes:         (1),(1,40,40,1)
        - Mean values:  Not specified
        - Scale values:         Not specified
        - Scale factor:         Not specified
        - Precision of IR:      FP32
        - Enable fusing:        True
        - Enable grouped convolutions fusing:   True
        - Move mean values to preprocess section:       False
        - Reverse input channels:       False
TensorFlow specific parameters:
        - Input model in text protobuf format:  False
        - Path to model dump for TensorBoard:   None
        - List of shared libraries with TensorFlow custom layers implementation:        None
        - Update the configuration file with input/output node names:   None
        - Use configuration file used to generate the model with Object Detection API:  None
        - Operations to offload:        None
        - Patterns to offload:  None
        - Use the config file:  None
Model Optimizer version:        2019.1.0-341-gc9b66a2
[ ERROR ]  Can't permute attrs for node EltwiseReshapeNormalization. Error message: only integer scalar arrays can be converted to a scalar index

/////////////////////////////////

Please, help me!!

0 Kudos
7 Replies
Shubha_R_Intel
Employee
224 Views

Dear park, kyungja,

I'm seeing  frozen_model_tf_1.14 everywhere but Model Optimizer only supports up to Tensorflow 1.12 today. 

Thanks,

shubha

park__kyungja
Beginner
224 Views

Dear, shubha,

Thank you for your quick reply.

Actually, I produced a frozen model on tensorflow 1.9 first

It gave me a same error message.

So, I upgraded tensorflow 1.14 and produced a frozen model on it.

Is any possible other reason?

Thanks anyway 

/////////////////////////

// my model info

C:\Program Files (x86)\IntelSWTools\openvino_2019.1.133\deployment_tools\model_optimizer\mo\utils>python summarize_graph.py --input_model "d:\openVINO\frozen_model_tf_1.9.pb"
2 input(s) detected:
Name: keep_probabilty, type: float32, shape: <unknown>
Name: input_image, type: float32, shape: (-1,40,40,1)
1 output(s) detected:
inference/prediction

/////////////////////////

Shubha_R_Intel
Employee
224 Views

Dear park, kyungja,

Can you tell me the kind of Tensorflow model you're having issues with ? Is it an Object Detection API model ? SSD ? Faster RCNN ? If it's a Tensorflow Object Detection API model, then you must follow the below instructions:

http://docs.openvinotoolkit.org/latest/_docs_MO_DG_prepare_model_convert_model_tf_specific_Convert_O...

Please report back here with your results.

Thanks,

Shubha

park__kyungja
Beginner
224 Views

Dear, Shubha

My model is a kind of semantic segmentation.

It's a modified model of FCN.

It works fine in Ubuntu system with NVIDIA GPU.

Thanks.

Kyungja

Shubha_R_Intel
Employee
224 Views

Dear park, kyungja,

Please run downloader.py under C:\Program Files (x86)\IntelSWTools\openvino_2019.1.148\deployment_tools\tools\model_downloader . Do python3 downloader.py --all . After completing that you will see a few semantic segmentation models that we also support, cascaded_unet, deeplab, dilation. Is the model you're trying one of these ?

If it's not, please give me a link to your model (I assume that it's a public model).

Thanks,

Shubha

 

park__kyungja
Beginner
224 Views

Dear Shubha,

I've already tested public semantic segmentation models.

There are no problem.

I want to have application specific model.

I designed my network and learned the model with my own image sets.

I can't give you my model publicly.

But, I can send it to you by email.

Please, let me know your email address.

 

Thanks.

Kyungja

Shubha_R_Intel
Employee
224 Views

Dear park, kyungja,

I have PM'd you so that you may send me the model privately.

Appreciate your cooperation.

Thanks !

Shubha

 

Reply