search

yonghenglh6 / DepthwiseConvolution

A personal depthwise convolution layer implementation on caffe by liuhao.(only GPU)
525 stars 184 forks source link

Did somesone train some models with this DepthwiseConvolution layer? #19

Open Ai-is-light opened 6 years ago

Ai-is-light commented 6 years ago

@yonghenglh6 Thanks for your good work! I trained caffe model by using your DepthwiseCovolution layer as following your suggestion, while it works when I train the model, I met some problems . When I deploy the model I trained , there was a error : Check failed registry count (type) == 1 (0 VS 1) Unknown layer type: DepthwiseConvolution. So , did you meet this error? or Have you deployed some caffe model with DepthwiseConvolution? Looking forward your attention. Thanks

yonghenglh6 commented 6 years ago

Which phase do you meet the problem in? It seems that your caffe did not merge the new layers. Please assure the .cpp and .h be placed in right position, and re-make has been done.

408550969 commented 6 years ago

Hi,I find the speed is down when I add DepthwiseConvolution layer. The original like this type:conv kernel:3 pad:1 stride:1 and I change like this: type:conv kernel:1 pad:0 stride:1 BN type:DepthwiseConvolution kernel:3 pad:1 stride:1

The speed is half the original,do you know why does this happen?

JoyLuo commented 6 years ago

@Ai-is-light Hi，the same question trouble me, could you solve it ?

ShawKai666 commented 6 years ago

@JoyLuo You must add them in your caffe.sln and re-make!

LyraWang commented 5 years ago

Hi,I find the speed is down when I add DepthwiseConvolution layer. The original like this type:conv kernel:3 pad:1 stride:1 and I change like this: type:conv kernel:1 pad:0 stride:1 BN type:DepthwiseConvolution kernel:3 pad:1 stride:1

The speed is half the original,do you know why does this happen?

I got the same problem as you, have you found out the reason？

408550969 commented 5 years ago

Sorry, I don't know. I finally gave up DepthwiseConvolution on caffe.