Thank you for this inspiring project—it’s been incredibly helpful!
I noticed that the same data (TestVolumes and TestSegmentation) is used for both selecting the best model during training and evaluating it in testing. Could this lead to overfitting, as the model indirectly "sees" the test data during selection? Wouldn't this affect its generalization ability?
Thank you for this inspiring project—it’s been incredibly helpful!
I noticed that the same data (TestVolumes and TestSegmentation) is used for both selecting the best model during training and evaluating it in testing. Could this lead to overfitting, as the model indirectly "sees" the test data during selection? Wouldn't this affect its generalization ability?