farmingyard / caffe-mobilenet_v2

caffe based mobilenet v2 deploy

caffe-mobilenet_v2

caffe based mobilenet v2 deploy, the depthwise conv, please refer to（https://github.com/farmingyard/caffe-mobilenet ）

caffe-mobilenet_v2's Issues

how to implement relu6

I do not understand how to implement relu6 operation that mentionde in paper,

and i do not found the ops in the the deploy.prototxt .can you tell me?
@farmingyard

segmentation fault

Any idea for solving this? I just re-compiled caffe after merging your dw_conv.
This shows when I use caffe timer on both cpu and gpu mode.

*** Aborted at 1520306614 (unix time) try "date -d @1520306614" if you are using GNU date ***
PC: @ 0x7f8694ae4b06 caffe::ConvolutionDepthwiseLayer<>::LayerSetUp()
*** SIGSEGV (@0x0) received by PID 31868 (TID 0x7f8695694ac0) from PID 0; stack trace: ***
@ 0x7f869372a4b0 (unknown)
@ 0x7f8694ae4b06 caffe::ConvolutionDepthwiseLayer<>::LayerSetUp()
@ 0x7f8694ca29a5 caffe::Net<>::Init()
@ 0x7f8694ca4a88 caffe::Net<>::Net()
@ 0x40bc2a time()
@ 0x407360 main
@ 0x7f8693715830 __libc_start_main
@ 0x407a19 _start
@ 0x0 (unknown)
Segmentation fault (core dumped)

Need pretrained caffemodel

@farmingyard
Do you have pretrained caffemodel of caffe-mobilenet_v2，Thank you!

About some problem

Thanks for you job!
After I looked at the prototxt you wrote, the comparison paper found several differences.
First,in res7_conv1 layer, the input is 28×28×32, I think channel of the layer is 32×6，but your prototxt is 144.And the same problem occurs in the res14_conv1 layer. Can you tell me the reason? Second, I believe you wrote the prototxt according to the input column in Table 2 of the paper, ignored s column, however, the paper wrote that mobilenet v2 has 19 bottlenect layers rather 17. how do you think?
Last, can you tell me the result of your training?
Thanks again !

Imagenet experiment result and pretrained model

Thanks for your network prototxt. May you provide Imagenet experiment result and pretrained model? I think it will be helpful.