Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.

plugin.get_supported_layers

yu__jin
Beginner
730 Views

HI~

I  installed the OpenVINO 2020.1 and I built the demo successfully and also run well for MYRIAD (use -d MYRIAD).

I have got my own onnx model and IR file(xml, bin).

Then I run the python script like this:

--------------------------------------------------------------

from openvino.inference_engine import IENetwork, IEPlugin
plugin=IEPlugin(device='MYRIAD')
testxml="/home/ryz/Pictures/siamfc.xml"
testbin="/home/ryz/Pictures/siamfc.bin"
net=IENetwork(model=testxml, weights=testbin)
sup_layers=plugin.get_supported_layers(net)
no_sup = [i for i in net.layers.keys() if i not in sup_layers]
print(no_sup)

-------------------------------------------------------------

then I got this result:     ['actual_input_1']

the layer-actual_input_1  in xml file like this:

        <layer id="0" name="actual_input_1" type="Parameter" version="opset1">
            <data element_type="f16" shape="1,3,127,127"/>
            <output>
                <port id="0" precision="FP16">
                    <dim>1</dim>
                    <dim>3</dim>
                    <dim>127</dim>
                    <dim>127</dim>
                </port>
            </output>
        </layer>

------------------------------

 I am confused... so what's wrong with this layer?

 

Best regards.

 

 

0 Kudos
4 Replies
yu__jin
Beginner
730 Views

And when i change the 

plugin=IEPlugin(device='MYRIAD')

to plugin=IEPlugin(device='CPU').

All layers are supported.

0 Kudos
SuryaPSC_Intel
Employee
730 Views

Hi Jin,

Can you share what type of model you are using?

Best Regards,

Surya

 

0 Kudos
yu__jin
Beginner
730 Views

Chauhan, Surya Pratap Singh (Intel) wrote:

Hi Jin,

Can you share what type of model you are using?

Best Regards,

Surya

 

I convert the siamfc pytorch model to onnx. It's a target tracking model.

Some operations in siamfc are not supported by openvino.

so when I convert the onnx to IR, I do the Model Cutting.

python3 mo.py --input_model siamfc.onnx --input=actual_input_1 --output=46

I don't konw what's wrong with actual_input_1 layer

       <layer id="0" name="actual_input_1" type="Parameter" version="opset1">
            <data element_type="f16" shape="1,3,127,127"/>
            <output>
                <port id="0" precision="FP16">
                    <dim>1</dim>
                    <dim>3</dim>
                    <dim>127</dim>
                    <dim>127</dim>
                </port>
            </output>
        </layer>

 

 

 

 

 

0 Kudos
SuryaPSC_Intel
Employee
730 Views

Hi Jin,

Siamese network is not a supported pytorch model via onnx conversion.

Also, layers support vary from device to device. Kindly refer to Supported Layers.

Best Regards,

Surya

0 Kudos
Reply