Undefined function 'extractHOGFeatures' for input arguments of type 'uint8'

Question

I am trying to extract HOG features from an image in MATLAB R2013a.

This is the code I copied from MATLAB documentation website:

input='1.png';
Z=imread(input);
%Z=rgb2gray(Z);
[featureVector, hogVisualization] = extractHOGFeatures(Z);

The error I get is:

Undefined function 'extractHOGFeatures' for input arguments of type 'uint8'.

Error in hogfeatureextractor (line 6) [featureVector, hogVisualization] = extractHOGFeatures(Z);

I think this is due to the computer vision toolbox not installed properly or it might be some other reason that I am not aware of. I have installed MATLAB under Ubuntu 14.04.

I have also used ver to determine if I have the computer vision toolbox installed and I do:

MATLAB Version: 8.1.0.604 (R2013a)
MATLAB License Number: 724504
Operating System: Linux 3.13.0-44-generic #73-Ubuntu SMP Tue Dec 16 00:22:43 UTC 2014 x86_64
Java Version: Java 1.6.0_17-b04 with Sun Microsystems Inc. Java HotSpot(TM) 64-Bit Server VM mixed mode
-------------------------------------------------------------------------------------------------------
MATLAB                                                Version 8.1        (R2013a)
Simulink                                              Version 8.1        (R2013a)
Aerospace Blockset                                    Version 3.11       (R2013a)
Aerospace Toolbox                                     Version 2.11       (R2013a)
Bioinformatics Toolbox                                Version 4.3        (R2013a)
Communications System Toolbox                         Version 5.4        (R2013a)
Computer Vision System Toolbox                        Version 5.2        (R2013a)

As such, I'm not sure why I'm experiencing this error. Can someone help me out?

Use another function for calculating HOG features. I recommend the one in Piotr Dollar's toolbox. — Autonomous
@ParagS.Chandakkar - Good choice. Another would be VLFeat - vlfeat.org/overview/hog.html — rayryeng
@rayryeng Piotr Dollar toolbox HOG function is fast, about 125 frames per second, 640 by 480 size. Even faster than OpenCV when I last checked about 6 months ago. I don't know about VLFeat HOG's speed. — Autonomous
@ParagS.Chandakkar - OK, but I didn't reference OpenCV. I referenced VLFeat. How fast is it compared to VLFeat? — rayryeng
Ya, probably I should have said "fast", not faster. My mistake ;). — Autonomous

rayryeng rayryeng · Accepted Answer · 2015-03-24T18:54:27

extractHOGFeatures was introduced in MATLAB R2013b. As you are running R2013a, you unfortunately do not have access to this function. See the release notes for the Computer Vision Toolbox here: http://www.mathworks.com/help/vision/release-notes.html and look under R2013b.

You have two options. One is to upgrade your version of MATLAB so that it's at least R2013b, or you can use a third party toolbox. Parag in his comment to you recommended Piotr Dollar's toolbox: https://github.com/pdollar/toolbox

If I can recommend something, try using VLFeat: http://www.vlfeat.org. VLFeat is an open source computer vision library that is implemented in C, and also MATLAB with MEX wrappers.

Check out the HOG tutorial of VLFeat for more details: http://www.vlfeat.org/overview/hog.html.

Undefined function 'extractHOGFeatures' for input arguments of type 'uint8'

1 Answers