Ask HN: What face recognition framework should I use?

6 points by mea36 15 years ago | 5 comments

I have an idea for an app I'd like to build that compares similar faces (I'd rather not say much more than that about the app) and I was wondering if anyone has advice about which face recognition framework I should use.

I know I can just google this and get a bunch of frameworks, but I'd like your advice.

ajray 15 years ago |

OpenCV is my favorite Computer Vision (CV) library. http://opencv.willowgarage.com/wiki/

Face detection is a special case of Haar-wavelets, which are essiantially recognitions that can be 'trained' by reinforcing with positive and negative images.

OpenCV comes with some really well pre-trained Haar-wavelets for recognizing faces. http://opencv.willowgarage.com/wiki/FaceDetection

mea36 15 years ago | |

Cool, thank you.

JonathanWCurd 15 years ago |

Open CV is free and decent to start with. Lots of wrappers exist for it so you could pretty much use it with whatever environment you are using. If your idea is validated you could then look into some of the more expensive commercial libraries out there. They would give you more features flexibility scale and (maybe) better results.

yemkay 15 years ago |

http://face.com (if you can afford API calls)

mea36 15 years ago | |

This is really cool. I may start with this and then move to OpenCV. The thing about this is that I want to be able to rate how the algorithm did so I can constantly improve the algorithm. Here the algorithm seems to be abstracted.