How can I deal or track only one face while I have detected all the faces in front of camera?

I have face recognition code which recognizes the face and show its name when I will generate the database for that, it is working well when only one face is in front of camera but I am using "FaceDetect = vision.CascadeObjectDetector;" for face detection which will detect all the faces in front of camera and I am getting bounding box around all the faces. Is there any way to track face of my choice among the detected all faces? Or can I code each bounding box face separately? Please give me some guideline. Thanks

Answer 1

[Aurele Turnes]

Once you have created your vision.CascadeObjectDetector object, you can apply it to your image I to detect the faces on the image as follows:

bboxes = step(faceDetect, I);

The returned matrix bboxes is M-by-4, where M is the number of faces detected and the 4 elements describe the bounding box around each face, as explained in the following documentation page:

http://www.mathworks.com/help/releases/R2014a/vision/ref/vision.cascadeobjectdetector.step.html

To access the ith face, you can simply extract the corresponding pixels from the original image:

i = 1;
indexi = bboxes(i,:);
facei = I(indexi(2):indexi(2)+indexi(4),indexi(1):indexi(1)+indexi(3),:);
figure;
imagesc(facei)

You can then pass facei to your face recognition algorithm to detect which person this face corresponds to and track who is on the picture.

[talal]

Thanks brother for this very useful information. Actually I am working on real time object tracking so I want to ask you that how will it work for live video capturing?

[talal]

Dear Turnes,

First of all thanks for your answer. I am following your guidelines but I am facing some problems can you please help me.

index1=bbox(1,:);
     face1=temp(index1(2):index1(2)+index1(4),index1(1):index1(1)+index1(3),:);
G=rgb2gray(face1);
R=imresize(G,[100,100]);
%     imagesc(face1)
name1=recognize(R);            .....my face recognition function

and I am getting error

Error using rgb2gray>parse_inputs (line 81)
MAP must be a m x 3 array.

Error in rgb2gray (line 35)
X = parse_inputs(varargin{:});

Error in get_face (line 11)
 img=rgb2gray(img);

Thanks in advance

[Aurele Turnes]

It seems like you are getting an error when you use the rgb2gray function because your frame face1 is only a 2D-matrix. The function rgb2gray expects its input to be in color, and thus to be a 3D-matrix. Can you check if your face1 frame is a 3D-matrix?

Answer 2

[Image Analyst]

Try Brett's work: http://www.mathworks.com/matlabcentral/fileexchange/46645-face-recognition-using-aggregated-features