Object Dectection Notebook: The operation, 'image_tensor', does not exist in the graph.

Question

Object Dectection Notebook: The operation, 'image_tensor', does not exist in the graph.

hemaAI opened this issue 5 years ago · comments

I followed your whole tutorial and it worked until I tried to reference my frozenmodel in the notebook. In the last step the output is: ""The name 'image_tensor:0' refers to a Tensor which does not exist. The operation, 'image_tensor', does not exist in the graph.""

Can you help?

yassineSaid · Answer 1 · Mon Feb 17 2020 04:49:03 GMT+0800 (China Standard Time)

Hello @hemaAI, did you found a solution? I have also the same issue

hema · Answer 2 · Mon Feb 17 2020 15:37:03 GMT+0800 (China Standard Time)

I think it was another tensor name. You have to print out your tensor names.
In my case it was something like input_tensor instead of image tensor.

yassineSaid · Answer 3 · Mon Feb 17 2020 17:11:59 GMT+0800 (China Standard Time)

When i displayed all tensors names, i found one with "input" as name but i used it i got this error:
Cannot feed value of shape (1, 576, 768, 3) for Tensor 'input:0', which has shape '(?, 299, 299, 3)'

hema · Answer 4 · Mon Feb 17 2020 17:16:15 GMT+0800 (China Standard Time)

thats exactly where I ended up and did not find a solution...

Gilbert Tanner · Answer 5 · Wed Jun 24 2020 22:50:34 GMT+0800 (China Standard Time)

Sorry for not replying. It seems like there are multiple things that can trigger this issue.

For some people replacing:
image_tensor = tf.get_default_graph().get_tensor_by_name('image_tensor:0')
with:
image_tensor = 'image_tensor:0'
solved the problem.

I will close this issue for now. If anyone experiences any further issues feel free to reopen it.