MATLAB: Does the current onnx version support the export of object detection networks, such as the Yolov2 network（export to yolov2.onnx）

deep learningDeep Learning Toolbox

Does the current onnx version support the export of object detection networks(use exportONNXNetwork function), such as the Yolov2 network（export to yolov2.onnx）? It seems that only support 'classification' ,'regression','pixelclassification' layer type network…

Best Answer

upcoming release R2020a will support exporting YOLO v2 object detectors to ONNX.

For R2019b, refers to the link below to check which layers are supported to export to ONNX.

https://www.mathworks.com/help/deeplearning/ref/exportonnxnetwork.html

Related Solutions

MATLAB: Does the function ‘exportONNXNetwork’ reports warnings about incompatibility while trying to convert the pre-trained yolov3 to onnx

Currently, MATLAB export support is limited to opset version 9, but the parameters used in the 'resize2Dlayer '(featureResize2) of 'yolov3ObjectDetector' requires export support of opset >=10. As a workaround, you can replace the 'featureResize2' with a new 'resize2Dlayer' that can be exported even with opset version 9. Here is an example:

lgraph= layerGraph(detector.Network);
layer = resize2dLayer('OutputSize',[28,28],'Name','resize2DLayer','GeometricTransformMode','asymmetric','NearestRoundingMode','onnx-10');
lgraph = replaceLayer(lgraph,'featureResize2',layer);
net = dlnetwork(lgraph);

The above net is a yolov3 network that can be exported without warnings.

To verify the mAP, you can create a detector as below and follow the steps mentioned in Evaluate Model section of the "Object Detection Using YOLO v3 Deep Learning" example.

yolov3Detector = yolov3ObjectDetector(net, detector.ClassNames, detector.AnchorBoxes);

MATLAB: Faster RCNN classifier to ONNX format

The “exportONNXNetwork(net,filename)” could be used to export the deep learning network net with weights to the ONNX format, but this function does not support all the deep learning layers. If you export a network that contains a layer that the ONNX format does not support, then exportONNXNetwork saves a placeholder ONNX operator in place of the unsupported layer and returns a warning. You cannot import an ONNX network with a placeholder operator into other deep learning frameworks. “RegionProposalLayer” in RCNN is not supported by the exportONNXNetwork function thus it will keep a place holder operator at that position, hence you cannot use it in other deep learning frameworks .However you can export a YOLO Object Detection Network to ONNX model format.

Refer to the below links for more information:

ExportONNXNetwork: https://www.mathworks.com/help/deeplearning/ref/exportonnxnetwork.html#d120e36378

Export YOLO v2 Object Detector to ONNX: https://www.mathworks.com/help/vision/examples/export-yolo-v2-object-detector-to-onnx.html

Best Answer

Related Solutions

MATLAB: Does the function ‘exportONNXNetwork’ reports warnings about incompatibility while trying to convert the pre-trained yolov3 to onnx

MATLAB: Faster RCNN classifier to ONNX format

Related Question