jocpae / clDice

MIT License
217 stars 28 forks source link

What does each index of [:,1:,:,:,:] mean? #6

Closed Darcy103 closed 3 years ago

Darcy103 commented 3 years ago

Can you explain what each index represents? My task is a segmentation task of two classifications (foreground and background), and the output of the network is 4-dimensional (batchsize, 2, H, W). I want to try the loss function proposed in the paper, but I haven't figured out why it is 5-dimensional.

jocpae commented 3 years ago

Hi, thank you for your comment, we have changed the notation to [:,1:,...], as all the dimensions after the foreground denote the spatial dimensions. (I asuume your case involves 2D data, making the 5th dimension redundant.) On why we only use the foreground please refer to our paper and a previous issue.