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
Thanks for your sharing, and it helped me a lot.
But I have a little question about the implementation of channel adjustment.
In 'shuffleNet_V2_1x_deploy.prototxt', the output channel number of resx1_match_DWconv is 58.
However, it seems that the input channel number of this layer is 24, which is really wired for a depthwise convolution.
Could you please explain this for me?
Thank you so much!!!
The text was updated successfully, but these errors were encountered:
Thanks for your sharing, and it helped me a lot.
But I have a little question about the implementation of channel adjustment.
In 'shuffleNet_V2_1x_deploy.prototxt', the output channel number of resx1_match_DWconv is 58.
However, it seems that the input channel number of this layer is 24, which is really wired for a depthwise convolution.
Could you please explain this for me?
Thank you so much!!!
The text was updated successfully, but these errors were encountered: