Conversion of tensorflow(.pb) model to intel Openvino IR (.bin and .xml)

[yr__rakesh_arya]

Hi ,

I am trying to convert Tensorflow model to IR.

During the conversion process i get this error. as shown in the attachment 

[ ERROR ]  Shape [ -1 224 224   3] is not fully defined for output 0 of "input". Use --input_shape with positive integers to override model input shapes.

[ ERROR ]  Cannot infer shapes or values for node "input".

[ ERROR ]  Not all output shapes were inferred or fully defined for node "input".

For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #40.

[ ERROR ]

[ ERROR ]  It can happen due to bug in custom shape infer function <function tf_placeholder_ext.<locals>.<lambda> at 0x000001920A07F9D8>.

[ ERROR ]  Or because the node inputs have incorrect values/shapes.

[ ERROR ]  Or because input shapes are incorrect (embedded to the model or passed via --input_shape).

[ ERROR ]  Run Model Optimizer with --log_level=DEBUG for more information.

[ ERROR ]  Stopped shape/value propagation at "input" node.

I trying to convert mobilenet_v1_1.0_224 model .it is standard tensorflow model example.

Please help me to under stand this issue  and what other parameters is required in the cmd whiling converting to IR ?

[Hyodo__Katsuya]

For example...

$ sudo python3 mo_tf.py \
--input_model xxx.pb \
--output_dir path/to/your/output \
--input input \
--output zzz \
--data_type FP32 \
--batch 1

[Kulkarni__Vinay]

Hi Hyodo,

      I did like this on windows machine :

Here python is python 3.6.5

C:\Users\Ignitarium\Documents\tensorflow-yolo-v3>python  C:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\model_optimizer\mo_tf.py --input_model yolo_v3.pb --tensorflow_use_custom_operations_config  yolo_v3_changed.json

I am getting errors as indicated above.

Also attached the python and .pb files also.

I am not really clear on what you are trying to say by example ?

C:\Users\Ignitarium\Documents\tensorflow-yolo-v3>python  C:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\model_optimizer\mo_tf.py --input_model yolo_v3.pb --tensorflow_use_custom_operations_config  yolo_v3_changed.json
Model Optimizer arguments:
Common parameters:
        - Path to the Input Model:      C:\Users\Ignitarium\Documents\tensorflow-yolo-v3\yolo_v3.pb
        - Path for generated IR:        C:\Users\Ignitarium\Documents\tensorflow-yolo-v3\.
        - IR output name:       yolo_v3
        - Log level:    ERROR
        - Batch:        Not specified, inherited from the model
        - Input layers:         Not specified, inherited from the model
        - Output layers:        Not specified, inherited from the model
        - Input shapes:         Not specified, inherited from the model
        - 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
        - Offload unsupported operations:       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:  C:\Users\Ignitarium\Documents\tensorflow-yolo-v3\yolo_v3_changed.json
Model Optimizer version:        1.4.292.6ef7232d
[ ERROR ]  Shape [ -1 416 416   3] is not fully defined for output 0 of "Placeholder". Use --input_shape with positive integers to override model input shapes.
[ ERROR ]  Cannot infer shapes or values for node "Placeholder".
[ ERROR ]  Not all output shapes were inferred or fully defined for node "Placeholder".
 For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #40.
[ ERROR ]
[ ERROR ]  It can happen due to bug in custom shape infer function <function tf_placeholder_ext.<locals>.<lambda> at 0x0000027A804BEAE8>.
[ ERROR ]  Or because the node inputs have incorrect values/shapes.
[ ERROR ]  Or because input shapes are incorrect (embedded to the model or passed via --input_shape).
[ ERROR ]  Run Model Optimizer with --log_level=DEBUG for more information.
[ ERROR ]  Stopped shape/value propagation at "Placeholder" node.
 For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #38.
 

[Hyodo__Katsuya]

python  C:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\model_optimizer\mo_tf.py --input_model yolo_v3.pb --tensorflow_use_custom_operations_config  yolo_v3_changed.json --batch 1

or

python  C:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\model_optimizer\mo_tf.py --input_model yolo_v3.pb --tensorflow_use_custom_operations_config  yolo_v3_changed.json --input_shape [1,416,416,3]

[Kulkarni__Vinay]

Thanks Hyodo, this helped.

[yr__rakesh_arya]

Thanks Hydo I need more help on this.

First i ran this cmd as told in the intel Openvino Documentation.

python mo.py  --input_model C:\Users\raryax\Desktop\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224_frozen.pb

i have used mobilenet_v1_1.0_224 model .it is standard tensorflow model .
after seeing the error (complete error log is below) it looks like there is some problem with shapes parameter 
so what value has to be passed to it  ? 

I ran this below cmd by passing --input_shape [1,416,416,3]
then i got this error

python mo.py  --input_model C:\Users\raryax\Desktop\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224_frozen.pb --input_shape [1,416,416,3]
 
 
mo.py: error: unrecognized arguments: --input_shape[1,416,416,3]

Model Optimizer arguments:
Common parameters:
        - Path to the Input Model:      C:\Users\raryax\Desktop\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224_frozen.pb
        - Path for generated IR:        C:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\model_optimizer\.
        - IR output name:       mobilenet_v1_1.0_224_frozen
        - Log level:    ERROR
        - Batch:        Not specified, inherited from the model
        - Input layers:         Not specified, inherited from the model
        - Output layers:        Not specified, inherited from the model
        - Input shapes:         Not specified, inherited from the model
        - 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
        - Offload unsupported operations:       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:        1.4.292.6ef7232d
[ ERROR ]  Shape [ -1 224 224   3] is not fully defined for output 0 of "input". Use --input_shape with positive integers to override model input shapes.
[ ERROR ]  Cannot infer shapes or values for node "input".
[ ERROR ]  Not all output shapes were inferred or fully defined for node "input".
 For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #40.
[ ERROR ]
[ ERROR ]  It can happen due to bug in custom shape infer function <function tf_placeholder_ext.<locals>.<lambda> at 0x000002B903FAF9D8>.
[ ERROR ]  Or because the node inputs have incorrect values/shapes.
[ ERROR ]  Or because input shapes are incorrect (embedded to the model or passed via --input_shape).
[ ERROR ]  Run Model Optimizer with --log_level=DEBUG for more information.
[ ERROR ]  Stopped shape/value propagation at "input" node.
 For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #38.
 
 
 please can explain me in details what value has to be passed for shape or should i pass more parameters  ? 
 
 
 
 

[Hyodo__Katsuya]

@yr, rakesh arya

>[ ERROR ]  Shape [ -1 224 224   3] is not fully defined for output 0 of "input". Use --input_shape with positive integers to override model input shapes.

Please try below.

python mo.py  --input_model C:\Users\raryax\Desktop\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224\mobilenet_v1_1.0_224_frozen.pb 
--input_shape [1,224,224,3]

[KLim4]

-

[Gdeep]

Dear yr, rakesh arya

try this command to convert the ssd-mobile-net model

python3 <mo_tf.py> --input_model <path_to .pb> --tensorflow_use_custom_operations_config </extension/front/tf/ssd_support.json> --input_shape [1,224,224,3]

[yr__rakesh_arya]

@Hyodo Thanks for the help it is working 

[li__heju]

hello,Hyodo:

 I've settled it.

Thanks;

[Lee__Hanbeen]

i have used a Ubuntu LTS 16.04

in my case,

sudo python3 mo_tf.py

--input_model <.pb file directory>

--input_shape "[1, 64, 128, 3]"

--input "input"         

--data_type FP32

it works for me    

[Agg__Áron]

Hi!

I had the same issue, is used this:

python3 mo_tf.py --input_model yolo_v3.pb --tensorflow_use_custom_operations_config yolo_v3.json --data_type FP16 --input_shape [1,416,416,3]

than i become another error:

E0701 13:40:49.641717 140493334865728 main.py:323] -------------------------------------------------
E0701 13:40:49.641915 140493334865728 main.py:324] ----------------- INTERNAL ERROR ----------------
E0701 13:40:49.641994 140493334865728 main.py:325] Unexpected exception happened.
E0701 13:40:49.642042 140493334865728 main.py:326] Please contact Model Optimizer developers and forward the following information:
E0701 13:40:49.642104 140493334865728 main.py:327] 
E0701 13:40:49.642600 140493334865728 main.py:328] Traceback (most recent call last):
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/main.py", line 312, in main
    return driver(argv)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/main.py", line 263, in driver
    is_binary=not argv.input_model_is_text)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/pipeline/tf.py", line 141, in tf2nx
    graph_clean_up_tf(graph)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/passes/eliminate.py", line 186, in graph_clean_up_tf
    graph_clean_up(graph, ['TFCustomSubgraphCall', 'Shape'])
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/passes/eliminate.py", line 181, in graph_clean_up
    add_constant_operations(graph)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/passes/eliminate.py", line 145, in add_constant_operations
    Const(graph, dict(value=node.value, shape=np.array(node.value.shape))).create_node_with_data(data_nodes=node)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/ops/op.py", line 207, in create_node_with_data
    [np.array_equal(old_data_value[id], data_node.value) for id, data_node in enumerate(data_nodes)])
AssertionError

E0701 13:40:49.642694 140493334865728 main.py:329] ---------------- END OF BUG REPORT --------------
E0701 13:40:49.642745 140493334865728 main.py:330] -------------------------------------------------

Hyodo, Katsuya can you give some solution? 

Thanks szkudo

[RANI__SOMYA]

I am trying to convert faster_rcnn_resnet50 frozen_inference_graph.pb to IR but unable to do it.

 python /opt/intel/openvino/deployment_tools/model_optimizer/mo.py --input_model /home/iquantela/Desktop/Video_Analytics/video-analytics-master/road_cleaning/tensorflow-1/models/research/object_detection/inference_graph/frozen_inference_graph.pb --output  detection_boxes,detection_scores,num_detections --tensorflow_use_custom_operations_config /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/faster_rcnn_support_api_v1.10.json --input_shape [1,1024,600,3]
Model Optimizer arguments:
Common parameters:
    - Path to the Input Model:     /home/iquantela/Desktop/Video_Analytics/video-analytics-master/road_cleaning/tensorflow-1/models/research/object_detection/inference_graph/frozen_inference_graph.pb
    - Path for generated IR:     /home/iquantela/.
    - IR output name:     frozen_inference_graph
    - Log level:     ERROR
    - Batch:     Not specified, inherited from the model
    - Input layers:     Not specified, inherited from the model
    - Output layers:     detection_boxes,detection_scores,num_detections
    - Input shapes:     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:     /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/faster_rcnn_support_api_v1.10.json
Model Optimizer version:     2019.1.1-83-g28dfbfd
 

[ ERROR ]  Input shape "1" cannot be parsed.