How can I extract the features of a data set?

Hi, I am trying to do a face recognition from web cam project and helping from this link: https://www.mathworks.com/matlabcentral/fileexchange/53849-code-for-face-recognition-with-matlab-webinar

So, There are first HOG feature extraction from a dataset and use its training features and training labels in machine learning command fitcecoc. When I use HOG features for extraction from data set, there is one row matrix for each image and the total feature matrix is a 320x4680 matrix for 320 images. But I want to use different extraction method. I tried to detect features and after that extract features. Then, there was one 2by2 matrix for each image and I couldn't find how I can compound these features in one feature matrix to put into machine learning command. So, how can I extract the features of this data set with different method to take a better solution? Also, I didn't understand how the command fitcecoc work. Is there any chance to use another method to do comparison between input image and data set? Thank you.