Hi all,
Currently I am doing face recognition using eigenfaces and neural network. In neural network part, I am using feedforward backpropagation and i have so much done with my feature extraction in eigenfaces.
The problems of mine is I stuck in the process of creating neural network. From matlab help,to create feedforward neural I have to type:
net = newff(p,t,10,{},'trainlm');
I am not sure what I should put as my target,t.
In my case, I have 20training images with 2images per person in training folder and 10testing images with 1 image person in testing folder. I would like to train a NN first and then test it with an image from testing folder to see whether the NN recognise the testing image anot.
I knew that my inputs,p should have 20 vectors because I have 20images in training folder. My target,t in creating the NN should be test image but how could I enter it first while i will not know which testing image will be insert into the system to be tested?