Error when i use NCS2 in Instance Segmentation Python* Demo

[wenju__niu]

when i use NCS2 in Instance Segmentation Python* Demo i got errors as follows

Traceback (most recent call last):
  File "instance_segmentation_demo/instance_segmentation_demo.py", line 279, in <module>
    sys.exit(main() or 0)
  File "instance_segmentation_demo/instance_segmentation_demo.py", line 154, in main
    exec_net = ie.load_network(network=net, device_name=args.device, num_requests=2)
  File "ie_api.pyx", line 85, in openvino.inference_engine.ie_api.IECore.load_network
  File "ie_api.pyx", line 92, in openvino.inference_engine.ie_api.IECore.load_network
RuntimeError: [VPU] Cannot convert layer "647" due to unsupported layer type "ExperimentalDetectronPriorGridGenerator"

Environment:

Ubuntu 18.04 x86_64 (Corei7 Gen8)

OpenVINO 2019 R3.1

[JesusE_Intel]

Hi niu,

Thanks for reaching out! I tested the python segmentation demo with the deeplab v3 model and did not see any issues. Which model are you using? Could you share a link to the model and the model optimizer command used to convert it to IR format?

Regards,

Jesus

[Tenn__Hian-Kun]

Hi,

I can confirm the error encountered by niu.

It was an instance segmentation model. According the error message, it might be caused by some unsupported layers of MYRIAD devices?

What Jesus E. used is a semantic segmentation model, right? AFAIK, it's different from the one used in instance segmentation, and maybe it has no unsupported layers?

[JesusE_Intel]

Hi Hian-Kun,

You are correct, I tested the semantic segmentation model/demo not the instance segmentation. The instance segmentation models are only supported on CPU. 

Regards,

Jesus