Community
cancel
Showing results for 
Search instead for 
Did you mean: 
idata
Community Manager
445 Views

How does NCS support caffe custom layers ?

How can I use NCS to convert and implement my own caffe layers ?

 

Or is there a way to use host's CPU to execute the layers ?
Tags (1)
0 Kudos
6 Replies
idata
Community Manager
147 Views

@cuihao1993 Thanks for your interest. Currently there isn't support for custom layers in the current NCSDK (v 2.05) and we cannot provide a road map or eta for this feature right now.

idata
Community Manager
147 Views

@Tome_at_Intel

 

Thank you for your kind answer !

 

May I use the cpu(s) inside the NCS to calculate some Intermediate result between two layers ?

 

Many thanks !
idata
Community Manager
147 Views

@Tome_at_Intel

idata
Community Manager
147 Views

@cuihao1993 Maybe you could achieve your goal by using mvNCCompile and the -in and -on options. You could compile a model with -in input and then use -on <output node> to specify the output node. This would result in compiling a graph file that takes input from the input node and its output node would be the specified output node. So instead of processing the data through the entire model, you can get the output of the intermediate results of whatever node you choose.

 

You can view the nodes in your model using the mvNCProfile tool. More information on the mvNCProfile tool can be found at: https://movidius.github.io/ncsdk/tools/profile.html

idata
Community Manager
147 Views

hello, @Tome_at_Intel .

 

And I have another question , in

 

https://movidius.github.io/ncsdk/caffe.html#vision

 

Regular Convolution - 1x1s1, 3x3s1, 5x5s1, 7x7s1, 7x7s2, 7x7s4

 

Does this mean that 3*3s2 and 5x5s2 is not supported?

idata
Community Manager
147 Views

I use mvNCProfile to get the runtime of a deploy.prototxt which contains conv 5x5s2 conv 3x3s2 successful.

 

So I think it support 3×3s2 and 5x5s2 . Am I correct?

 

Thank you !
Reply