Utilize VPU of Myriad-X in NCS2 Stick

[mfeu]

Hi,

I purchased several Intel NCS2 Sticks and I can run inference on different neural networks on it.

However, the product page state, that the Myriad-X chip not only has an integrated neural compute engine but also 16 programmable 128-bit VLIW Vector Processor optimized for vision workloads and also enhanced vision accelerators (see here: https://www.intel.com/content/www/us/en/products/docs/processors/movidius-vpu/myriad-x-product-brief.html)

I am wondering if and how they can be utilized on the NCS2 stick. In particular, I would be interested to run an object detection network on the stick, cropping the found object, and feeding it to a classifier on the same stick without copying results from and to the stick. Is this possible to do?

Where can I find documentation on how to utilize the whole feature set of the Myriad-X chip not only the inference part which is covered by the OpenVino Toolkit?

Looking forward to your answers!

[Wan_Intel]

Hi Mfeu,

Thanks for reaching out to us.

You can use Object Detection Python Demo from Open Model Zoo with Supported Models using Intel® NCS2 with the following command:

object_detection_demo.py -m <path_to_model> -i <input> -at <architecture_type> --labels "path_to_labels_mapping_file" -r -d MYRIAD

Next, you may refer to How to crop an image in OpenCV using Python for the steps to crop the found object after running an object detection network.

For your information, Intel® NCS2 is a plug and play development kit that can be used on low-power applications. It offers you access to neural network functionality, without the need for large, expensive hardware.

However, to develop complex solutions, we recommend using Intel® Vision Accelerator Design With Intel® Movidius™ Vision Processing Unit (VPU).

On another note, to access the documentation for Intel® Movidius™ Myriad™ X Vision Processing Unit, you have to register for Intel Premier Access here.

Regards,

Wan

[Wan_Intel]

Hi Mfeu,

Thanks for your question.

This thread will no longer be monitored since we have provided suggestions. 

If you need any additional information from Intel, please submit a new question.

Best regards,

Wan