Basic Face Detection and Face Recognition Using OpenCV

Detect faces and their eyes, nose and mouth in a web cam feed. Then calculate the rotation of each face. Next you can also capture the faces and store them in a format suitable for recognition. I then demonstrate working face recognition.

This is done using OpenCV, the code is open-source and you can get it at:
https://code.google.com/p/visual-control/

If you use it I would be happy to receive a notification of that. If you modify it, you are very welcome to commit your changes!

Look at my website for a basic post of the framework: http://toefel.nl/projects/computer-vision-framework/

If you have trouble compiling OpenCV with visual studio 2010, look at:
http://www.youtube.com/watch?v=XeBhwbRoKvk

For information about compiling Boost with visual studio 2010, look at:
http://www.youtube.com/watch?v=5AmwIwedTCM

I found the information and explanations about face detection, face tracking and face recognition from Robin Hewitt very useful:
http://www.cognotics.com/opencv/servo_2007_series/index.html

Category:

Howto & Style

Tags:

• Visual
• Control
• Face
• Detection
• Recognition
• Eyes
• Mouth
• Nose
• Rotation
• OpenCV
• Boost
• Eigenface
• Eigenfaces
• haar
• classifier
• technology
• principal
• component
• analysis

[87sgeric]

hi! i tested the .exe , why does it crash every time i tried saving a face? is there a fix directory I have to placed it in?

[toefel18]

@87sgeric make sure the extension of the image is .bmp (without the extension, the application might crash)

[dhruvmini01]

Thanks for the application. I have a question though. is there a way i can take frames from the eye, nose, and mouth detectors? Could you tell me how if possible

[toefel18]

@dhruvmini01 the system provides frames to the detectors, they analyze it for features. The class DetailedFaceDetector first scans for faces, if it finds a face, it tries to find the eyes, nose and mouth. Basic detectors expose their findings trough the getAreas() method, but DetailedFaceDetector provides a getDetailedFaceInfo() method which returns a map that maps faces to a feature map, it looks like: (facerectangle => (facefeature=>featurerectangle­)) you can look it up in google code.

[toefel18]

@TheRikkiJSmith i replied a while ago

[rebkahibasdeys]

hey, i downloaded the zip source code given on the website.

can i modify the code to track, lets say a cup or a ball....

best.

if you could help me with that i'd be great...

[toefel18]

@rebkahibasdeys Hey, you can use the visual-control source project at google code and write a new detector class that contains the code for detecting a ball, cup, or whatever. You can look at the FaceDetector class to see how I implemented the face detection. I used the HaarLike features detector, but that's slow for tracking a ball. I am sure you can find much more efficient methods for that!, good luck.

[87sgeric]

@87sgeric nvm i solved it, i have to fill in the image extension. Thanks!!

[TheRikkiJSmith]

could you get back to that email I sent you about openCV and my final year project thanks :)

[toefel18]

@rebkahibasdeys you can modify the source code yourself, this app is meant to get ppl started with a working demo!

[rebkahibasdeys]

third query, any way i can save the output of this application??