Null coefficients when classifying an image

Hello guys! I have a question related to image classification problems.

After training my network and finishing the offline phase and saved trained network.

Trained network : mynet= trainNetwork(augmentedTrainingSet, lgraph, opts);

I want to count the number of null coefficients of the network (unused weights) when classifying the image in the online phase to make a comparision and determine the reason of time gain between many networks.

%% Make a prediction using the classifier

[YPred,probs] = classify (mynet, newimage);

So after making this prediction in the online phase, how can I get the connections' weights used in the classification of this new image?

Thank you! Appreciate any help!