YOLOv4 exportONNX Warnings

Hi, I am trying to export the YOLOv4 "tiny-yolov4-coco" network as ONNX using the exportONNX function. It is currently giving me the following warnings:

Warning: ONNX does not support layer 'nnet.cnn.layer.FunctionLayer'. Exporting to ONNX
operator 'com.MathWorks.Placeholder'.
Warning: ONNX does not support layer 'nnet.cnn.layer.FunctionLayer'. Exporting to ONNX
operator 'com.MathWorks.Placeholder'. 
Warning: ONNX does not support layer 'nnet.cnn.layer.FunctionLayer'. Exporting to ONNX
operator 'com.MathWorks.Placeholder'.
Warning: Cannot exactly export Layer 'up2d_35_new' with 'GeometricTransformMode' set to
'half-pixel' for 'OpsetVersion' '8'. Exported network may produce different results. 
Warning: Cannot exactly export Layer 'up2d_35_new' with 'NearestRoundingMode' set to 'round'
for 'OpsetVersion' '8'. Exported network may produce different results. 

I realise that the functions here are just not supported yet on exportONNX based on the documentation. Are there alternatives to the above functions? Also, are there applications that can edit these placeholder values as ONNX files?

[Sivylla Paraskevopoulou]

I don't know how to replace placeholders in ONNX. But if exporting to TensorFlow works for your workflow, in R2022b MATLAB introduced the exportNetworkToTensorFlow function. If the MATLAB network or layer graph contains a custom or built-in MATLAB layer that exportNetworkToTensorFlow cannot convert to a TensorFlow layer, the exportNetworkToTensorFlow function exports this layer as a custom TensorFlow layer. Here is an example on how to deal with exported custom TensorFlow layers: Export Layer Graph with Custom Layer to TensorFlow.

[Erick Cardozo]

I have the same problem. Staff matlab Can You solve this?. The problem is the layer slice (function layer), this function layer is not compatible with onnx. yolov3 and yoliv4 are basic models for Deep Learning, it would be a shame if they could be exported to onnx.