How to calculate total variation (TV) prior on a 4D cudatensor?

6 views

Skip to first unread message

aMY

unread,

Oct 7, 2017, 10:22:51 AM10/7/17

to torch7

I have two tensors whose dimensions are: (batchsize, channels, ht, wt). Can anyone suggest how to calculate the total variation on the batch of images denoted by the 4D cudatensor? Should I access each element of the tensor and compute the prior or is there a better method?

Reply all

Reply to author

Forward

0 new messages