Are those keras loss and accuracy weird?

Question

I have a relatively small mri dataset and I'm trying to do a binary segmentation. I have built an ordinary U-Net structure and trained it.

But the output seems a bit weird to me. Both train and validation accuracies stucked at a value first, but then both accuracies made a sudden big jump at 27th or 28th epoch.

Loss graph looks more acceptable, next is the graphs:

Accuracy Graph:

Loss Graph:

I have another issue that even if I have an %97-98 accuracy on training data, when I tested it on some images from training data, results converted to binary mask were not that good.

Then I have decreased the threshold from 0.5 to 0.35 while retrieving output images and the results were almost perfect.

What do you think about that? thanks in advance.

Answer 1

They seem a little off with those stuck epochs, it really means the model isn't learning (weights are not changing, new cases are not providing useful information) but that is totally plausible.

Just to be sure. What optimizer are you using? and did you try with another one?