12

In my application I'm going to implement an image search like Google image search. Only difference is here I'm searching only for human faces. Here's how it goes..

  • user take a picture of a person using the camera.
  • my app should identify the face of that person and get details from the image regarding the face (eg: colors).
  • Using those details app will do a search in the image database for similar faces.
  • Finally it'll display results to the user.

So I want to recognize faces using Android face recognition class and select that recognized area and save the details of that face into an array or a database.

I got few questions on above scenario.

  • what are the things I need from that image to do a proper image search?
  • Can I do it only using colors?
  • Do you have any idea on getting pixel details from that recognized face? I know how to do it on bitmap images. But i'm not sure that method still works with the face recognition class.

Can anyone explain those thing to me please?

ahmednabil88
  • 13,263
  • 9
  • 45
  • 80
Ramesh Jaya
  • 641
  • 1
  • 10
  • 16

2 Answers2

5

Use OpenCV's Recognition API.

I wrote a bit about how to set it up in another question.

Community
  • 1
  • 1
Brian Attwell
  • 8,683
  • 2
  • 29
  • 26
  • 1
    isn't it available in JAVA so that i can use it in My Android Project ??? Can you be more specific to Android Link of Open CV – Dr. aNdRO Dec 16 '13 at 07:19
3

I recommend you too to use OpenCV functions, they are well optimized for face detection and recognition .

for recognition, you don't need to colour processing. Since you develop your application on Android, you can use OpenCV4Android

Here there some helpful links, that let you have a good idea to start:

Face Recognition on Android

Here you find a useful tutorial for face detection and recognition

http://www.shervinemami.info/faceRecognition.html

I also recommend you to apply the detection api that already implemented in android stuff:

http://pi-virtualworld.blogspot.com.br/2013/04/android-opencv-face-detection.html

In order to prepare a clean and a proper face database, which in turn enhance the goodness of recognition results, don' forget to align faces http://bytefish.de/blog/aligning_face_images/

Try and test already benchmark face database, to evaluate your implementation:

http://vis-www.cs.umass.edu/lfw/#deepfunnel-anchor

http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html

Community
  • 1
  • 1
Y.AL
  • 1,813
  • 13
  • 27