Image segmentation submission

After successfully creating the model, I cannot upload tot model.h5 in the submission column. It always cuts off or gets stuck at a point. Any suggesstion ?

You filed this under DLS Course 3, but there are no assignments in C3. It sounds like you are talking about Week 3 of Course 4. But if I have that right, then I don’t understand your question. There is no point at which you need to upload anything. You just click the “Submit Assignment” button, right?

It is week 3 of course 3. I’m suppose to submit the model.h5 file after the training but It is not uploading.

Hi @Benjamin_Appiah_Yebo

DLS course3 is ‘Structure Machine Learning Project’ with only 2 weeks of the course material, there is no week3. In addition, there is no lab assignment, only quizzes for this course.

Hi @Benjamin_Appiah_Yebo,

You have posted this in the Deep Learning Specialisation, but I have a feeling you are talking about a course in one of the Tensorflow specialisations.

Can you provide us with the name of the course this is from ?


Tensorflow specialization course 3 Advance Computer vision

Tensorflow specialization course 3 Advance Computer vision. That’s the name of the course. I made a mistake my bad.

No worries, @Benjamin_Appiah_Yebo, I have moved your topic to TF-AT Course 3 Week 3 for you.


1 Like

No worries. But be careful about this in the future, because posting something in the wrong place will not get you the help in time.

For now, let me ask someone to help you out with this,

Hi @tf3_mentors,

Can one you help @Benjamin_Appiah_Yebo here ?


@gent.spah gave some good suggestions on the other thread where you asked about this: Image segmentation-Submission problem - #2 by gent.spah

1 Like

hi Wendy

I’ve tried the suggestions, but the file is still failing to get uploaded. Could it be because of the file size (over 28M)?


That large file size could well be a problem. There’s an easy way to fix that:

  1. Kernel → Restart and Clear Output
  2. Save
  3. Submit

Of course that means you lose your training results but that also happens if the notebook just “times out”. So you’ll have to rerun later. But the grader literally does not need any of the generated output: just the code.

Well, to be a bit more precise: you don’t lose the ability to see the training results once the notebook times out, but you can’t actually use them. You can only visually examine them. Once the kernel times out, none of the runtime data is actually usable and you have rerun everything. But you can see the output …

Thanks Paul for the quick response. I’ve managed to train a new model with size smaller than 20M and now the file uploaded fine.

Thanks again