search

hellloxiaotian / ESRGCNN

Image Super-resolution with An Enhanced Group Convolutional Neural Network (Neural Networks, 2022)
https://github.com/hellloxiaotian/ESRGCNN
60 stars 5 forks source link

conv2d groups=1 #5

Closed iperov closed 1 year ago

iperov commented 2 years ago

groups=1 is default value and changes nothing from regular Conv2d()

hellloxiaotian commented 2 years ago

Dear Mr or Mrs,

This paper uses different channels to divide different group to mine relations of different channels in terms of depth and width to extract more robust features in image super-resolution.

Best wishes, Chunwei

At 2022-06-25 19:57:05, "iperov" @.***> wrote:

groups=1 is default value and changes nothing from regular Conv2d()

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>