How to train a group of images Using SVM classifier ?

Question

syscon on 5 Mar 2014

0
Link

Direct link to this question

https://se.mathworks.com/matlabcentral/answers/120255-how-to-train-a-group-of-images-using-svm-classifier

Commented: Walter Roberson on 13 Aug 2018

I downloaded some codes for GENDER IDENTIFICATION on color images Using classifier But It always shows an Error "mkl_custom.dll" not found.

Kindly help me how to Train a group of images Using SVM(nonlinear) classifier for gender identification.

0 Comments
Show -2 older commentsHide -2 older comments

Sign in to comment.

Sign in to answer this question.

Answer 1

priyanka gadade on 17 Feb 2016

1
Link

Direct link to this answer

https://se.mathworks.com/matlabcentral/answers/120255-how-to-train-a-group-of-images-using-svm-classifier#answer_210141

what the procedure to label the training data ? Labels convert feature data into binary form.. plz tell me procedure for Feature to label conversion.

0 Comments
Show -2 older commentsHide -2 older comments

Sign in to comment.

Answer 2

Anand on 6 Mar 2014

0
Link

Direct link to this answer

https://se.mathworks.com/matlabcentral/answers/120255-how-to-train-a-group-of-images-using-svm-classifier#answer_127321

I don't know anything about the code you're using, but if you would like to use an SVM classifier, look at this

Support Vector Machines in MATLAB

4 Comments
Show 2 older commentsHide 2 older comments

Salma Hassan on 22 Jul 2017

sir your comment exactly solve my problem can you help me with code link . that i have a feature vectors and the next step is how to make a representation of this features with machine learning

Walter Roberson on 22 Jul 2017

As Has:

See http://www.mathworks.com/help/stats/fitcecoc.html#examples

Note: you cannot use binary SVM for this task, as over 50 genders have been described so far.

Estimates for the portion of humans who do not fit into 20th century "western civilization" binary notions of gender vary quite a bit, from 0.03% to 1% to 11%. However, since this task involves classification from photographs rather than actual gender, you need to take into account a different statistic, which has to do with the portion of individuals whose body has features not typically associated with traditional binary gender; the figure for that with regards to those who would typically be considered "male bodied" has been estimated to be as high as about 58% in one autopsy study.

Sign in to comment.

Answer 3

syscon on 17 Mar 2014

0
Link

Direct link to this answer

https://se.mathworks.com/matlabcentral/answers/120255-how-to-train-a-group-of-images-using-svm-classifier#answer_128798

when i install with "custom settings" deselecting some other toolbox..it shows error like that .

but now i installed full package.there no errors like that. obviously dont know to classify the images with svm classifier.

1 Comment
Show -1 older commentsHide -1 older comments

Anand on 21 Mar 2014

Why is this a separate Answer?

What "custom settings" are you talking about? Which package are you referring to?

Sign in to comment.

Answer 4

Walter Roberson on 21 Jun 2016

0
Link

Direct link to this answer

https://se.mathworks.com/matlabcentral/answers/120255-how-to-train-a-group-of-images-using-svm-classifier#answer_226218

Trying to identify gender by image is the modern day phrenology, except with even less science behind it. Gender is internal, and every known biochemical or physical measurement to test gender objectively has been found to be incorrect. At best there are correlations. And as for correlations, I highly recommend reading https://www.thestar.com/news/insight/2016/01/16/when-us-air-force-discovered-the-flaw-of-averages.html