2

I'm doing a university project for my final year of a Computer Science course, and it involves facial recognition as students are walking into a classroom. I'm trying to find an SDK that can process multiple images (if I have a camera set up that can take 1 image per second it should get every student, however some testing does need to be done and documented)

Students do not come into a classroom in a line as such, so there may be more than one face on an image but if I place the camera in the right place they should all be facing forwards in at least one of the images - I then need to find the identity of each face using a database which has pictures.

This is where I'm lost, a lot of facial recogniton API's can only cope with one person/image, or just detect that there is somebody there but has no "compare" implemented so I can't use the database.

Ideally it needs to free, since the university would like to deploy it if I complete it successfully, does anybody have any ideas what I can use?

PS. Similar to the kind of thing they use in movies/TV shows I suppose...where they just run something over an image, get the faces then compare them to a national database. Not sure if something like that exists in real life, or which I can use on a smaller database, but I hope so.

  • Dude if this was a project why would it be just implementable with an API? Which language are you using ? Have you checked out any API's – cjds Dec 14 '12 at 03:17
  • Because I'm not expected to do all the math/low level stuff and create a solution myself. I'm expected to use something that already exists, the brief gave face.com as an example, which died years ago... I have tried biometrycloud.com as an alternative, no-one got back to me... and betaface but it doesn't provide the identity recognition I need. –  Dec 14 '12 at 03:32
  • Okay I have a question. In one camera shot. Do all the students come? Are multiple faces seen? – cjds Dec 14 '12 at 03:34
  • In one camera shot you would get maybe 3-4 faces, there will be multiple shots will eventually get every student. Then it can be worked out using either a web API or a c# library, I just need to know what library/API's are out there. –  Dec 14 '12 at 03:41
  • Check the answer. +1 if you like it. – cjds Dec 14 '12 at 03:45
  • Hi! Thank you very much, I'm looking at them now and they seem to be good leads.... however the system won't let me +1 you without 15 rep :( I'm sorry... –  Dec 14 '12 at 04:00

2 Answers2

3

Here are a list of the API's you can use:

Sky Biometry

HP Analytics Beta

Both of them should solve the problem of multiple people. Personally I like Sky Biometry and I think the API is better.

Check out this

Community
  • 1
  • 1
cjds
  • 8,268
  • 10
  • 49
  • 84
0

OpenCV would suit your requirements. It has a rich image processing API and can help you detect multiple faces at a time, recognize them, recognize gestures, etc. It provides you programs with which you can "train" OpenCV to recognize faces, objects, also it is Open Source. There is a lot of info on this, just type in OpenCV and search. A couple of links to get you going :) :

The OpenCV Website

A Good site with tutorials on using OpenCV

stamhaney
  • 1,246
  • 9
  • 18