You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Caffe Users
Hi all. I'm trying to implement a network in Caffe that requires uneven padding, e.g. adding just one zero row or col to the side, rather than one on each side. This is easily done in tensorflow and pytorch, but I see that Caffe's convolution and padding layers don't natively support this; does anybody know of a hacky workaround just to get the desired image size? Thanks!
赵Chris
unread,
Jul 10, 2018, 6:10:30 AM7/10/18
Reply to author
Sign in to reply to author
Forward
Sign in to forward
Delete
You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Caffe Users
You can modify the caffe.proto and let the pad_h, pad_w become pad_top, pad_bottom, pad_left and pad_right, then use these params in conv layer calculations.
It's a tedious and error-prone work.
在 2018年7月10日星期二 UTC+8上午8:01:19,Josh Gruenstein写道:
Przemek D
unread,
Aug 22, 2018, 10:10:14 AM8/22/18
Reply to author
Sign in to reply to author
Forward
Sign in to forward
Delete
You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Caffe Users
Give us a little while more, there's a work in progress PR which adds this exact functionality.