You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When I read the paper, I was confused about how is the input feature map factorized into high-frequency and low-frequency branches? Or if the input feature map is just randomly split among channel dimensions, and then use the following operations to extract high-frequency and low-frequency information from Xh1, Xh2, and Xl?
Best,
Yulin
The text was updated successfully, but these errors were encountered:
Thank you for your wonderful job.
When I read the paper, I was confused about how is the input feature map factorized into high-frequency and low-frequency branches? Or if the input feature map is just randomly split among channel dimensions, and then use the following operations to extract high-frequency and low-frequency information from Xh1, Xh2, and Xl?
Best,
Yulin
The text was updated successfully, but these errors were encountered: