-
|
I am recently implementing yolo-face500k-v2(https://github.com/dog-qiuqiu/MobileNet-Yolo) with TF, and noticed that there is Thanks in advance. |
Beta Was this translation helpful? Give feedback.
Answered by
gitE0Z9
Jun 21, 2021
Replies: 1 comment
-
|
After a little search, I found this layer is also used in efficientnetb0 and how to implement this layer in TF. If no scale_wh provided, simply multiply input with indexed layer output, its function is to pay attention to more important pixel in SE block. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
gitE0Z9
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
After a little search, I found this layer is also used in efficientnetb0 and how to implement this layer in TF. If no scale_wh provided, simply multiply input with indexed layer output, its function is to pay attention to more important pixel in SE block.