Breast cancer detection model - Error with Keras Learner Node

Dear KNIME Community,

I am writing to inform you that I am currently working on a breast cancer detection project and would like to export the model to model.h5 in or
der to continue the project in Python. I have been using the Keras Node to create my model, but I have encountered an error while using the Keras Learner Node. The error message states “The input shape does not match the tensor shape. [1, 3] vs. [150, 150, 3].”

I have been unable to find a solution for converting the images, and I am also detecting three types of cancer. there’s no TO IMAGE conversion in the drop menu inside Keras Network Learner node.

I would greatly appreciate any guidance or suggestions that the KNIME community can provide on how to resolve this issue and export my model.




The workflow:
LUNGDetection.knwf (51.9 KB)

Thank you for your time and support.

Hi @AliGad,

could you try installing KNIME Image Processing - Deep Learning Extension – KNIME Community Hub ?
This should add a conversion method to allow for image input in the Keras Leaner node.

Let me know whether that worked!

Best regards
Steffen

1 Like

Hi @steffen_KNIME,
Thank you for your replay.
Actually I used “image reader” node from KNIME Image Processing - Deep Learning Extension – KNIME Community Hub to read training and testing images and it’s not allowed in the conversion method inside in the “Keras Leaner node”.

Hi,

we are talking about two different extensions. You installed KNIME Image Processing – KNIME Community Hub
Please try installing the other extension.

Steffen

2 Likes

Okay, I got it and will try it.
Thanks @steffen_KNIME

thank you @steffen_KNIME
IT WORKS!
Thank you @steffen_KNIME so much for your incredible support.