I'm trying to deploy to a google aiy vision kit v2 a custom model trained with Keras API in Tensorflow,when compiling the model from a .pb
Tensorflow file with the vision kit compiler i get:
output node named "<output node>" does not appear in the graph.
My question is, how can i get from tensorboard the output node passed to the compiler in the parameter --output_tensor_names
?
I already tried the values returned by:
keras_model.outputs[0].op.name # returns final_output/Sigmoid
keras_model.outputs[0].name # returns final_output/Sigmoid:0
and final_output
, which is the name given to the last layer in keras.
But none of these worked, how can i see the correct expected node in tensorboard?