medical-polar-training
medical-polar-training copied to clipboard
On using small target segmentation dataset
Hello, I intend to use the model to the small target segmentation task (as shown in Figure 1), but I found that the trained model saves diceloss=0 (Figure 2 below)
figure 1
figure 2
In the test, the dice score=0,the other three metrics are 1.(Figure 3 below)
Figure 3
I don't know what causes this, and I am very much looking forward to your reply.
Thank you!