Loading Pretrained 'EfficientDet' Network

7 次查看（过去 30 天）
Avivit
Avivit 2024-3-28
评论： Venu 2024-4-1
Hello, I was going through examples from the deep learning toolbox when I ran into an issue specifically with the 'EfficientDet-D0' model from https://github.com/matlab-deep-learning/pretrained-efficientdet-d0
I ran the following from the README.md at: https://github.com/matlab-deep-learning/pretrained-efficientdet-d0/blob/main/README.md
model = helper.downloadPretrainedEfficientDetD0;
From the above, I should be able to access the dlnetwork object 'net' from the model, but I get the following response:
In helper.downloadPretrainedEfficientDetD0 (line 24)
Warning: Unable to load instances of class nnet.onnx.layer.ElementwiseAffineLayer into a heterogeneous array. The definition of
nnet.onnx.layer.ElementwiseAffineLayer could be missing or contain an error. Default objects will be substituted.
Dot indexing is not supported for variables of this type.
Error in dlnetwork/isQuantized (line 1417)
b = this.PrivateNetwork.isQuantized();
Error in deep.internal.quantization.QuantizedNetworkDisplay/mixinQuantizedNetworkHeader (line 23)
strategy = net.getStrategy(isQuantized(net));
Error in dlnetwork/getHeader (line 1427)
header = this.mixinQuantizedNetworkHeader(header, className);
I don't see why dot indexing would be an issue here, is there anything I missed that could be causing this? I am using Matlab version 2023b .
Many thanks in advance.
  1 个评论
Venu
Venu 2024-4-1
HI @Avivit
Try reinstalling the Deep Learning Toolbox Converter for ONNX Model Format Support Package

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

prabhat kumar sharma
Hi Avivit,
I understand you're encountering issues with downloading the pretrained EfficientNetB0 model for object detection or transfer learning tasks in MATLAB. Here are steps to address the issue:
  1. Ensure your MATLAB version is R2020b or newer for compatibility.
  2. Verify the ONNX Support Package installation with:
if isempty(which('importONNXNetwork'))
disp('ONNX support package is not installed.');
else
disp('ONNX support package is installed.');
end
3.If not installed, add it via the Add-On Explorer in MATLAB or execute:
onnxsupportpackage.install
4.Attempt to load your model using the imagePretrainedNetwork function.
For further guidance, the following resources may be helpful:
https://www.mathworks.com/help/deeplearning/ref/efficientnetb0.html
https://www.mathworks.com/help/deeplearning/ug/classify-image-using-googlenet.html
https://www.mathworks.com/help/deeplearning/ref/efficientnetb0.html
I hope it helps!

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

产品


版本

R2023b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by