Porting custom keras model to openvino

[Biradar__Abhishek]

Hi All,

I am trying to port the keras model shared in the link to openvino.

I got keras model converted to tf model first and now trying to convert tf model to openvino by using mo_tf.py..  I used the command 

python3 "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo_tf.py" --input_model xyz.pb --input_shape "[1,32, 32, 3]" --data_type FP32

But I am not able to convert and I am getting the error as mentioned below.. Could guys please help me get this model converted to openvino?

	- Input layers: 	Not specified, inherited from the model
	- Output layers: 	Not specified, inherited from the model
	- Input shapes: 	[1,32, 32, 3]
	- 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: 	2019.3.0-408-gac8584cb7
[ ERROR ]  -------------------------------------------------
[ ERROR ]  ----------------- INTERNAL ERROR ----------------
[ ERROR ]  Unexpected exception happened.
[ ERROR ]  Please contact Model Optimizer developers and forward the following information:
[ ERROR ]  shapes (64,2) and (0,) not aligned: 2 (dim 1) != 0 (dim 0)
[ ERROR ]  Traceback (most recent call last):
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/main.py", line 298, in main
    return driver(argv)
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/main.py", line 247, in driver
    is_binary=not argv.input_model_is_text)
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/pipeline/tf.py", line 163, in tf2nx
    for_graph_and_each_sub_graph_recursively(graph, fuse_linear_ops)
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/middle/pattern_match.py", line 58, in for_graph_and_each_sub_graph_recursively
    func(graph)
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/middle/passes/fusing/fuse_linear_ops.py", line 267, in fuse_linear_ops
    is_fused = _fuse_add(graph, node, fuse_nodes, False)
  File "/opt/intel/openvino_2019.3.376/deployment_tools/model_optimizer/mo/middle/passes/fusing/fuse_linear_ops.py", line 206, in _fuse_add
    fuse_node.in_port(2).data.set_value(bias_value + np.dot(fuse_node.in_port(1).data.get_value(), value))
ValueError: shapes (64,2) and (0,) not aligned: 2 (dim 1) != 0 (dim 0)

[ ERROR ]  ---------------- END OF BUG REPORT --------------
[ ERROR ]  -------------------------------------------------

Thanks,

Abhishek

[SuryaPSC_Intel]

Hi Abhishek,

As we had a discussion in an earlier thread, you confirmed to have converted the model to IR. Kindly try converting the model to frozen tensorflow .pb and then use model optimizer as you did previously.

Best Regards,

Surya

[Biradar__Abhishek]

--Reserved

Thanks,

Abhishek

[SuryaPSC_Intel]

Hi Abhishek,

I generated the tensorflow model from keras using a script attached in this thread, and then I was able to generate the IR using model optimizer. Please find the command as well as logs below. 

C:\Program Files (x86)\IntelSWTools\openvino_2020.2.117\deployment_tools\model_optimizer>python mo_tf.py --input_model C:\Users\suryap1x\Desktop\liveness.model.pb --input_shape [1,32,32,3]
Model Optimizer arguments:
Common parameters:
        - Path to the Input Model:      C:\Users\suryap1x\Desktop\liveness.model.pb
        - Path for generated IR:        C:\Program Files (x86)\IntelSWTools\openvino_2020.2.117\deployment_tools\model_optimizer\.
        - IR output name:       liveness.model
        - 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:         [1,32,32,3]
        - 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
        - Use the config file:  None
Model Optimizer version:        2020.2.0-60-g0bc66e26ff

[ SUCCESS ] Generated IR version 10 model.
[ SUCCESS ] XML file: C:\Program Files (x86)\IntelSWTools\openvino_2020.2.117\deployment_tools\model_optimizer\.\liveness.model.xml
[ SUCCESS ] BIN file: C:\Program Files (x86)\IntelSWTools\openvino_2020.2.117\deployment_tools\model_optimizer\.\liveness.model.bin
[ SUCCESS ] Total execution time: 15.26 seconds.

Best Regards,

Surya

[Biradar__Abhishek]

Not sure why it crashed when I was trying to convert the model usinf 2019.R1, Anyway I will update the openvino toolkit and try again..

Thanks,

Abhishek