Closed Realdr4g0n closed 4 years ago
Hello,
The code you provided is group convolution. Our implementation is an extreme case of group convolution where the number of groups equals the number of channels. Our implementation is equivalent to figure 4 in the Xception paper.
Thanks to your reply
In your paper I've mistaken you for depthwise separable conv but it is not you propose a conditional depthwise convolution similar to group conv.
am I right?
In my paper we use depthwise separable conv. In your code, when group=nin it's depthwise separable conv (same as what is used in my paper); when group < nin, it is group conv.
Thanks to your explain
I understand what I want to know
Hello xh-liu
I have some question about your paper and code
When I read Xception(https://arxiv.org/abs/1610.02357) and find code about this code of depthwise is have more parameter(group) like follow code
Can you explain about why this code and your code(In the depthwise conv part) is diffrent?