Detect Face and Track in a Video in MATLAB

 Face detection is a process to find human faces in a digital image. There are various ways to do that and we can use different programming languages to do it. In this post we will see how we can detect face and track in a video in MATLAB.

It can be done in the following steps:

1. Read the video and find the face location using the computer vision face cascade 

% Create a cascade detector object.

faceDetector = vision.CascadeObjectDetector();

% Read a video frame and run the face detector.

videoReader = VideoReader('up2.mp4');

videoFrame   = readFrame(videoReader);

faceLocation = step(faceDetector, videoFrame);

2. Place the location vectors in the frame and display it using imshow

% Draw the returned bounding box around the detected face.

detectedFrame = insertShape(videoFrame, 'Rectangle', faceLocation);

figure; imshow(detectedFrame); title('Detected face');

%%

faceLocationPoints = bbox2points(faceLocation(1, :));

featurePoints = detectMinEigenFeatures(rgb2gray(detectedFrame), 'ROI', faceLocation);

pointTracker = vision.PointTracker('MaxBidirectionalError', 2);

% Initialize the tracker with the initial point locations and the initial

% video frame.

featurePoints = featurePoints.Location;

initialize(pointTracker, featurePoints, detectedFrame);

3. Create the video player object to play the video

videoPlayer  = vision.VideoPlayer('Position',...

    [100 100 [size(detectedFrame, 2), size(detectedFrame, 1)]+30]);

4. Run the following program in a loop to track the face through the frames

hile hasFrame(videoReader)

    % get the next frame

    videoFrame = readFrame(videoReader);

    % Track the points. Note that some points may be lost.

    [featurePoints, isFound] = step(pointTracker, videoFrame);

    newPoints = featurePoints(isFound, :);

    old_points = oldPoints(isFound, :);

    

    if size(newPoints, 1) >= 2 % need at least 2 points

        

        % Estimate the geometric transformation between the old points

        % and the new points and eliminate outliers

        [transformed_Rectangle, old_points, newPoints] = estimateGeometricTransform(...

            old_points, newPoints, 'similarity', 'MaxDistance', 4);

        

        % Apply the transformation to the bounding box points

        faceLocationPoints = transformPointsForward(transformed_Rectangle, faceLocationPoints);

                

        % Insert a bounding box around the object being tracked

        reshaped_Rectangle = reshape(faceLocationPoints', 1, []);

        detectedFrame = insertShape(videoFrame, 'Polygon', reshaped_Rectangle, ...

            'LineWidth', 2);

                

        % Display tracked points

        detectedFrame = insertMarker(detectedFrame, newPoints, '+', ...

            'Color', 'white');       

        

        % Reset the points

        oldPoints = newPoints;

        setPoints(pointTracker, oldPoints);        

    end

    

    % Display the annotated video frame using the video player object

    step(videoPlayer, detectedFrame);

end

5. The last thing to do is to release the video player object

Watch the video "Detect Face and Track in a Video in MATLAB" on Youtube

Download MATLAB code for face detection and tracking in a video

MATLAB Code