Why OpenCV face detection recognition the faces for untrained face?

asked 2017-04-05 03:26:38 -0600

Dhananjay Sharma gravatar image

I trained 472 unique images for a person A for Face Recognition using "haarcascade_frontalface_default.xml". While I am trying to detect face for the same person A for the same images which I have trained getting 20% to 80% confidence, that's fine for me.

But, I am also getting 20% to 80% confidence for person B which I have not included in training the images. Why its happening for person B while I am doing face detection?

I am using python 2.7 and OpenCV 3.2.0-dev version.

Please reply me.

edit retag flag offensive close merge delete

Comments

1

There are some inconsistencies in your question

  • face recognition <-> haarcascade : cascades are for face detection, which is localizing a face in a given image, from any person.
  • face recognition is matching an identity to a found face, which is done with the facerecognizer interface, found here

So you should make up what you want to do. My guess is the second thing!

StevenPuttemans gravatar imageStevenPuttemans ( 2017-04-05 04:30:13 -0600 )edit

Hello @StevenPuttemans

I have to Train the many people' images ( i.e. Dhananjay, Rajesh, Priyanka. etc.) using Python, OpenCV.

Now I want to Detect face and Recognition/Verification of the face from Trained data set. Where the given images are in Trained Data set or not?

Thanks,

Dhananjay Sharma gravatar imageDhananjay Sharma ( 2017-04-11 07:36:46 -0600 )edit
add a comment