Hi, I am trying to deploy a custom Neural Net on the NCS. I have two questions.
1) My custom NN has two inputs, an image and a separate scalar. As far as I can tell, the NCS can only accept one input. Is that correct?
2) What is the maximum number of parameters for a NN to still run on the NCS?
@mumes At the moment, we only support one input on the Intel NCS. We don't have a specific parameter limit for a NN, but there are some limitations. There are some unsupported TF operations and some graph size limitations (so this will inherently limit the size of possible networks that can run on the NCS). More information can be found in the errata list @ https://github.com/movidius/ncsdk/releases/tag/v1.12.00.01 and https://movidius.github.io/ncsdk/Caffe.html https://movidius.github.io/ncsdk/tf_compile_guidance.html.