Got error during run DL tensorflow example in KNIME(example 03_Train_MNIST_classifier)

Hi everyone,

I need some help.

I run the example name 03_Train_MNIST_classifier but got some errors. There is struck at the node "DL Network Executor"

The KNIME reported an error in the message below.

"WARN DL Network Executor 4:25 Selected input columns do not provide enough elements (1) to populate all neurons (784) of network input ‘input:0’. Try adding some columns to the selection."
WARN TensorFlow Network Writer 4:31 Output file ‘C:\Users\mypath\knime_temp_space_40971_30622\03_Train_MNIST_classifier30626\03_Train_MNIST_classifier\’ exists and will be overwritten

I click the noded (DL Network Executor) then select configure… In the Inputs section/Exclude Filter. There are no columns in the list (inside red rectangular box)

I’m not sure how I can fix this problem. My purpose just run to test the example and see the result.

Thank you very much for help.

I installed these following files

  • KNIME v3.7.2
  • Anaconda3 (tensorflow v.1.14 , pandas v.0.23.3, numpy v.1.16.4)
  • Python3.5
    In KNIME File/Preferences setting in KNIME/Python configuration: I set path to Python3 executable is > C:\Anaconda3\python.exe

Hi @cppong,

Did you install the “KNIME Image Processing - Deep Learning Extension” from the Community Contributions? This workflow needs this extension because it needs to know how to feed an image (from KNIME Image Processing) to a deep learning model. If this extension is installed the workflow (as configured on the Examples server) should work automatically.

Also, make sure to select the “Conversion” “From Image” for the input in the node configuration and the “Image” column is added to the “Include” list.



Hi @bwilhelm , I try what’s you recommended. It’s work fine. Now I can run the example w/o any errors.

Yeah… the problem was missing of “KNIME Image Processing - Deep Learning Extension" module.

Thank you very much. :grinning:


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.