So I just finished a prototyp for my thesis. Still need to segment the real data myself and collect some statistics stuff to write about the network, but I am pretty proud of the result considering the dataset is very small.

For now I need some god damn sleep.

Image semantic segmentation I guess

Looks like overfitting

@bioDan Why do you think the model is overfitting?

The dataset is very small and the results seem too accurate.

Im missing a lot of information to conclude its overfitting, but given only those two facts and the attached foto, i can confidently guess this is a case of overfitting, and it can easily be tested by presenting the model with new data and see how the prediction behaves.

@bioDan The attached image belongs to the validation dataset, which the model has never seen before. Modern segmentation approaches don't need that much data anymore to achieve a good result, but I get your point (Which also will take great importance in my thesis)