Use dice loss as the loss function,but sometimes loss is less than 0 about attention_unet

[MoleImg]

If you use my code of dice loss directly, you may find it is not the exact Dice loss according to its definition, so it might be possible to become negative. You can realize your own dice function according to your own task. Never mind.

[GusRoth]

Do you have any corresponding articles on these networks? @MoleImg

[MoleImg]

@GusRoth I did not realize the networks EXACTLY according to the specific articles. I just integrated the modules into my networks. I can provide you some as follows:

1. For the original U-Net: Ronneberger, Olaf, Philipp Fischer, and Thomas Brox. "U-net: Convolutional networks for biomedical image segmentation." International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015.
2. For the SE module: Hu, Jie, Li Shen, and Gang Sun. "Squeeze-and-excitation networks." Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
   Hope can help u well.

[BonesCat]

ok,thank you very much.I have fixed that problem.

[JokerWDL]

Do you have any corresponding articles on Spatial attention？thanks！