frame

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Does ncs support multi output layer in caffe or tensorflow now?

Comments

  • 4 Comments sorted by Votes Date Added
  • Hi @curry_best

    The MTCNN network is supported in the NCSDK and can be used for face detection.* But you would have to use version 1.12 of the NCSDK since there are still bugs being work on in the newer versions (supported in Tensorflow). I hope this information was helpful.

    *see NCSDK release notes here

    Best Regards,
    Sahira

  • Thanks @Sahira_at_Intel
    I know 1.12 version which use multiple NCS implement MTCNN for face detection. But 2.X version can load multiple graph in once and MTCNN has three graph. So has there some way to implement MTCNN with one NCS?

  • Hi @curry_best

    I think there might be, but you'd have modify the code in the ncappzoo to run on only one device. This is an old thread, but this NCS user posted a blog about how they used MTCNN to implement only one NCS. Let me look into this more and get back to you. Thanks for your patience.

    Sincerely,
    Sahira

  • Hi @Sahira_at_Intel
    You misunderstood what I meant. At the same time, that blog was written by me. I just want to know if NCS can handle multi output layer?

Sign In or Register to comment.