Home Discussion Colfax Cluster RuntimeError: Cannot make shared blob! The blob type cannot be used to store obj

This topic contains 4 replies, has 2 voices, and was last updated by  dhia 1 week, 6 days ago.

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #8762

    dhia
    Participant

    hello i’m trying to adapt the example of :safety gear on my own model faster rcnn tensorlow :
    i generated the IR without issues
    when i want to start the inference i get the error :
    RuntimeError: Cannot make shared blob! The blob type cannot be used to store objects of current precision
    in the line : exec_net,input_blob,out_blob = loadNetwork(plugin, net)

    #8763

    Ryo
    Keymaster

    Hi dhia,

    Unfortunately I am not familiar with this error. You might have a better luck asking at the Intel Developer Zone forums.

    With that said, based on the error message I would think that the issue is precision level used in the model. When you ran the model optimizer on your model, did you use FP16 or half? If so, can you try FP32 or full instead?

    Best,
    Ryo

    #8764

    dhia
    Participant

    Hello, i tried with the FP32 and i got another error instead :
    Primitive descriptor was not found for node FirstStageFeatureExtractor/resnet_v1_101/resnet_v1_101/block3/unit_23/bottleneck_v1/conv3/Conv2D.

    #8765

    Ryo
    Keymaster

    The error you are seeing sounds like you would need to create a custom layer extension for your model, as it is failing at extraction stage. You can find information on custom layers in the openvino docs.

    But once again, I recommend checking with the IDZ forums. As this is an Intel tool, they have experts there that can help you.

    – Ryo

    #8766

    dhia
    Participant

    Alright, thank you.

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.