MATLAB: How to classify 1000 images consisting of 10 classes of 100 images each(Wang database) using patternnet

Deep Learning Toolboxpatternnetwang

Hi! I have a collection of 1000 images consisting in 10 classes of 100 images each (Wang database) and I want to classify them using patternnet. In order of this, I extracted a feature vector of 1×64 dimensions for each image and I built a matrix of 1000×64. How to determine inputs and targets for patternnet?

Best Answer

 [ I N ] = size(x)  % [ 64 1000 ]
 [ O N ] = size(t)  % [ 10 1000 ]
where the columns of t are columns of the {0,1} unit matrix eye(10).
Hope this helps.
P.S. See the iris dataset example in the documentation
 help patternnet
 doc patternnet
 [ x, t] = iris_dataset;
  whos
 t = t    % No semicolon
Thank you for formally accepting my answer
Greg
Related Question