How can I use my CNN after training (Image Recognition)

4 visualizzazioni (ultimi 30 giorni)
Hello everyone,
I'm new to MATLAB but am working on a project so I've problem using and understanding new function and I'm not too good at deep learning with respect to hands on approach.
I've written code for CNN and trained the network somehow but I'm not able to use that trained network, will be greatful for any help.
Here's part of my code: -
%% Creating the CNN
% I've defined inputlayer, middlelayer and finallayer and it's training successfully
layers = [
inputlayer
middlelayer
finalayer
];
options = trainingOptions('sgdm','MaxEpochs',50, ...
'InitialLearnRate',0.000001);
convnet = trainNetwork(trainData,layers,options);
thisNetwork = convnet;
save('TheTrainedCNN','thisNetwork');
If I load my network with command using : -
load('TheTrainedCNN');
OR
load('TheTrainedCNN.mat')
I don'd get any error.
Then when I try to use it using: -
result = TheTrainedCNN('test.jpg');
I get error as:-
Unrecognized function or variable 'TheTrainedCNN'.
  1 Commento
Rik
Rik il 23 Mar 2021
@zayeema masoom Which thread do you think this is duplicating? And why didn't you respond to Adam?

Accedi per commentare.

Risposte (1)

Adam
Adam il 22 Gen 2020
When you saved the file the network was called 'thisNetwork'.
That is what this instruction saves:
save('TheTrainedCNN','thisNetwork');
'TheTrainedCNN' is the filename so when you load it as:
load('TheTrainedCNN');
your network will be deposited back into the workspace as 'thisNetwork', not 'TheTrainedCNN'.
  1 Commento
Adam
Adam il 23 Gen 2020
Like I said, your network is called thisNetwork. You should be able to see it in your workspace. I'm not quite sure how you would get from my answer that you should try:
result = convnet( 'test.jpg' )
and expect it to work!

Accedi per commentare.

Categorie

Scopri di più su Deep Learning Toolbox in Help Center e File Exchange

Community Treasure Hunt

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

Start Hunting!

Translated by