Train simple CNN

In this workflow we create a simple Convolutional Neural Network using the DL Python Network Creator. We train this network on our image data using the DL Python Network Learner and finally score it using the DL Python Network Executor. The DL Python Network Learner and Executor can be used to write custom training and execution code using Python.

This is a companion discussion topic for the original entry at

I have received an error like
‘DataFrame’ object has no attribute ‘convert_objects’

Hi @gsilah -

I was able to download and execute the workflow OK (although note that I also had to grab and execute its predecessor workflow, and download the associated data from Kaggle as described in its instructions).

What node is throwing the error for you? Can you list which version of KNIME you’re using, and on which OS?

1 Like

Hey @gsilah,

to fix this you need to install an older version of pandas that supports the function “convert_objects”.
To install a compatible pandas version in your anaconda environment, just run the following command:
conda install --name <your_env_name> pandas=0.20.3

I am having the same problem, I wish there’s a solution by changing the python code or updating it to the current pandas version compatible code, I am having all kinds of hassle trying to uninstall and install pandas 0.20.3, seems like my other packages are not compatible with it.

good day.
In process 01_Preprocess_image_data when reading images cats and dogs in a component in a node
Image Reader (Table)
errors, for example
“The number of cells in Row975 is not equal to the length of the array of column names: 1 vs 3”
(I use the kagglecatsanddogs_3367a set), Please tell me the set without errors or what is the problem.

1 Like

I have the same problem. The example workflow “Building CNN Predictor for cats and dogs images” does not work because it stops at the Image Table Reader with this error.

поменял набор данных на официальный с kaggle и проблема исчезла.