Closed ysh329 closed 7 years ago
@ysh329 Sorry. i just copy the README.md from caffe. I only give a pretrained model. I am also waiting for high speed depth-wise convolution
@KeyKy Thanks!
@KeyKy My big brother, do u have email or wechat. I'm to implement a high perforance depth-wise convolution :laughing:
@ysh329 370846270@qq.com wechat: xiaozhenyouni Nice!
I'm a little confused about
README.md
.You wrote:
This model is converted from MobileNet-Caffe
. However, I found the operator your implementedchannelwise_convolution.cc
,channelwise_convolution.cu
in mxnet@c492b5b. Besides, does thischannelwise
equal todepth-wise
convolution in original paper?