I AM WORKING ON PROJECT OF "FACE RECOGNITION FOR SECURITY" USING LABVIEW. I WANT TO KNOW HOW CAN I COMPARE AN IMAGE FROM CAMERA WITH DATASET OF IMAGES ALREADY EXISTING IN A TRAINING FOLDER?
First, check your CAPS LOCK key as it appears to be broken. Or maybe some food in it keeping it stuck on.
Second, your project is very ambitious, especially for a new LabVIEW user. Search the forums as questions like this come up from time to time and you may find some tips in them.
I'm almost certain you'll need to use the Vision toolkit. And probably need generate a lot of special algorithms to be able to compare a camera image to stored images as a way of evaluating similarity. Two pictures of the same person with the same hair style and clothes could wind up looking very different. Factor in different days, lighting conditions, hairstyles. glasses vs. none. makeup. I bet there could be huge differences between images of the same person.
Is this a school project?
I think modern algorithms involve face geometry.
Sounds good but I didn't get that. Is there any website or channel to understand this better?
Who are you asking?
If it is me you're asking (1st hit on google btw):
I have just released a couple of VIs for face detection and face recognition on VI package manager. The algorithms are based on convolutional neural networks. For face recognition the VI computes feature vectors for known faces and stores them in a file. For an unknown face the algorithm also computes a feature vector and computes a distance metric to the previously stored vectors. It will output the 5 best matches.
The VIs can be found on VIP by searching for 4SmartMachines.
In labview there are preimplemented techniques which are known as image processing algorithm blocks there we have to refine the the image and then we get the result in Labview for face detection my question is to please tell me what is the best image processing algorithm used in this technique
Trying to do face detection or face recognition with individual image processing operations is not state-of-the-art anymore.
The best results are achieved now with deep learning based approaches.
Face recognition is basically a two-stage process. In the first stage (face detection) possible candidate regions for faces in the image are found. In the second stage face (face recognition) decriptor vectors are calculated on these regions and these vectors are then used for recognizing or comparing faces.
Both of these stages are best done with deep learning approaches.
If you want to check the performance you can achieve with such a framework you can try this set of VIs: http://sine.ni.com/nips/cds/view/p/lang/de/nid/218050