- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi
Would it be possible to get the logic from mvNCCompile packaged as a shared or static library? It'd be nice to be able to convert networks in my binary, without having to have the mvNCCompile tool installed on the machine.
thanks
bill
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@billtohara, since you are asking for a shared library as against a namespace package, I presume you are developing your apps in C/C++ (not python); is that right?
without having to have the mvNCCompile tool installed on the machine
I can imagine a developer wanting to integrate training, testing, model conversion, and inference all into one single python script, but your request seems to be more from a deployment perspective. I am interested to learn more about your requirement, can you please give me some background to this ask?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Yes, developing in C++.
We have a binary we distribute that can work with caffe and tensorflow models. Ideally we'd be able to support NCS as an accelerator alongside GPUs (I've been asked to find non-nvidia options). However, for various reasons outside my control, the requirement is that we don't distribute python scripts etc.
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page