How to solve the error when trying to convert a DAGNetwork to ONNX?

I use Matlab R2021b and toolbox Converter for ONNX model version 21.2.

I only used these code:

net=load('pointnetplusTrained.mat','net').net;

filename = "pointnetplusTrained.onnx";

exportONNXNetwork(net,filename)

My net here has a value of 1X1 DAGNetwork and looks like this

The model is trained the same as Matlab documentation:

https://de.mathworks.com/help/releases/R2021b/lidar/ug/aerial-lidar-segmentation-using-pointnet-network.html

Error message shows below

Error using containers.Map/subsref

The specified key is not present in this container.

Error in nnet.internal.cnn.onnx.ConverterForConvolutionLayers/toOnnx (line 19)

inputTensorLayout = TensorLayoutMap(inputTensorNames{1});

Error in nnet.internal.cnn.onnx.ConverterForNetwork/networkToGraphProto (line 115)

= toOnnx(layerConverter, nodeProtos, TensorNameMap, TensorLayoutMap);

Error in nnet.internal.cnn.onnx.ConverterForNetwork/toOnnx (line 45)

modelProto.graph = networkToGraphProto(this);

Error in nnet.internal.cnn.onnx.exportONNXNetwork (line 17)

modelProto = toOnnx(converter);

Error in exportONNXNetwork (line 38)

nnet.internal.cnn.onnx.exportONNXNetwork(Network, filename, varargin{:});

Error in untitled3 (line 3)

exportONNXNetwork(net,filename)

Thanks in advance.

Answer 1

[Don Mathis]

Hi,

Unfortunately, this is a bug in the ONNX exporter. If you are able to edit your MATLAB source code, you can fix it as follows:

In MATLAB, enter

edit nnet.internal.cnn.onnx.ConverterForUnsupportedLayer

Then change line 42 from

    if isequal(OutputNames, {'out'})

to

    if numel(OutputNames)==1

[Hao Ding]

Thanks for the answer. The bug get fixed