I am trying to convert a Tensorflow model using the Model Optimizer in OpenVINO R4. Because the first few nodes are not supported by the model optimizer, I had to cut the model from the front and use port:node notation to specify the port. I also specify the input shape. The conversion works.
Now in addition to the input shape, I also want to specify mean_values. However, when I did this, Model Optimizer complained that it could not find the node I specified using the port:node notation. However, if I just specify the node, then it complains it has more than one inputs. What is the best way that I should work around this? Please see the attachments for the Tensorflow model I am converting, as well as the conversion script and the output.