Exception occurred during running replacer "REPLACEMENT_ID"

[UKR]

Hi, Im using a custom neural network developed in pytorch with two inputs named frame and prob. I converted the model to onnx format and trying to convert it to openvinoIR format. I get an error during the conversion. Can someone help me here or provide an example where conversion has been done for models with two inputs ?

Please find the configuration settings and the observed error below, 

Exporting ONNX model to IR... This may take a few minutes.
Model Optimizer arguments:
Common parameters:
	- Path to the Input Model: 	\openvino_notebooks\notebooks\102-pytorch-onnx-to-openvino\Segmentation_colors .onnx
	- Path for generated IR: 	\openvino_notebooks\notebooks\102-pytorch-onnx-to-openvino\.
	- IR output name: 	Segmentation_colors 
	- Log level: 	ERROR
	- Batch: 	Not specified, inherited from the model
	- Input layers: 	frame, prob
	- Output layers: 	Not specified, inherited from the model
	- Input shapes: 	[1,512, 896, 3], [1,512, 896, 4]
	- Mean values: 	Not specified
	- Scale values: 	Not specified
	- Scale factor: 	Not specified
	- Precision of IR: 	FP16
	- Enable fusing: 	True
	- Enable grouped convolutions fusing: 	True
	- Move mean values to preprocess section: 	None
	- Reverse input channels: 	False
ONNX specific parameters:
	- Inference Engine found in: 	\openvino_env\lib\site-packages\openvino
Inference Engine version: 	2021.4.0-3839-cd81789d294-releases/2021/4
Model Optimizer version: 	2021.4.0-3839-cd81789d294-releases/2021/4
[ ERROR ]  Exception occurred during running replacer "REPLACEMENT_ID" (<class 'extensions.front.user_data_repack.UserDataRepack'>): No node with name frame

[Zulkifli_Intel]

Hello Uthara Keerthan Ramamoorthy,

Greetings to you,

Can you share with us your model? We will take a look at it.

Sincerely,

Zulkifli

[UKR]

Yes, of course. 

Which file format do you need ? Can I send the onnx model format ?

[UKR]

Please find the onnx model. 

[Zulkifli_Intel]

Hello Uthara Keerthan Ramamoorthy,

Based on the Conversion Parameter documentation, conversion with 2 inputs is possible.

It seems like the input notes were not recognized. I successfully converted your ONNX model to IR format without using any parameters/arguments. Please give it a try and share with us the result.

Sincerely,

Zulkifli 

[Zulkifli_Intel · Accepted solution]

Hello Uthara Keerthan Ramamoorthy,

Here is another workaround you can try, for --input frame, prob try to change it to --input 0,1.

Sincerely,

Zulkifli

[UKR]

Yes, It worked. Thanks a lot for the response.

[Zulkifli_Intel]

Hello Uthara Keerthan Ramamoorthy,

Would you mind telling us which method did you used? Convert the ONNX model without the arguments or change the input nodes to 0,1?

Sincerely,

Zulkifli

[UKR]

Yes, I changed the input nodes to 0 and 1.

[Zulkifli_Intel]

Hello Uthara Keerthan Ramamoorthy,

Thank you for confirming your issue has been resolved. This thread will no longer be monitor. If you need any additional information from Intel, please submit a new question.