visualize network maskrcnn object and display weights

Hello,

I have an maskrcnn object - is there a way to visualize the NN behind it and give out the trained weights?

Cannot find a hint about it at the documentation:

https://de.mathworks.com/help/vision/ref/maskrcnn.html

Best Regards

Answer 1

[Manish]

Hi Odo,

I understand that you would like to visualize the neural network architecture of 'maskrcnn' and display its trained weights.

To visualize the neural network architecture of 'maskrcnn' , you can utilize a function called ‘analyzeNetwork’.

This function helps in visualizing the network's architecture. To find the trained weights of the network, you can iterate through each layer.

Here is the sample code for your reference:

maskobj= maskrcnn("resnet50-coco");
layers = maskobj.Layers;
analyzeNetwork(layers);
for i = 1:numel(layers)
    layer = layers(i);
    if isprop(layer, 'Weights')
        disp(['Layer ', num2str(i), ' Weights:']);
        disp(layer.Weights);
    end
    if isprop(layer, 'Bias')
        disp(['Layer ', num2str(i), ' Bias:']);
        disp(layer.Bias);
    end
end

Here is the documentation link for the function ‘analyzeNetwork’:

• https://www.mathworks.com/help/deeplearning/ref/analyzenetwork.html

hope this helps!