Face and Eye Tracking with LabView and Vision

[alfie90]

Hey all,

For my senior project I am trying to perform face tracking and eye tracking by using LabView and the Vision Development Module. Unfortunately I am new to LV and I'm drowning in a sea of links and tutorials so I was wondering if anyone here that has had experience performing similar tasks would be willing to help and point me in the right direction or provide examples. I bought two Logitech c525 webcams and hooked them up to my pc running LabView and Vision 2011. I was able to see images captured but that's as far as I've gotten with the tools. Any tips?

Thanks,

Alex.

[prasanth143]

am also doing same concept but for different application

[df86]

The simplest, and a very unreliable, way forward is to use Pattern Match. You might get erratic results though depending on the uniformity of the orientation/size/shape of the faces. Since it's a school project and you can control the parameters you might be able to get it to work. At the very least it will show you the difficulties you will have to overcome to make it more robust, as far as adjusting for grey scale shifts, size, etc. 

If you would like to actually put some effort and programming behind it below is a link which overviews the effort.

http://sine.ni.com/cs/app/doc/p/id/cs-14086 

Good luck, if you have any specific questions feel free to post and we'll try to walk you through it.

[MoviJOHN]

If you are familiar programming in C/C++, you can look at OpenCV.  There is plenty of code there that does this kind of thing.  To say you want to do this in LabVIEW, you will need to understand how head and eye tracking algorithms have been developed.  It is not trivial, to say the least.

[ZcuBa]

I have seen setups where IR diodes was placed below / above the screen.

The iris reflects the IR light, and the camera then captures the eyes as shining bright spots, almost saturated circles.

it was quite robust, and the eyes could be tracked for most scenarios, (depending on the relative angle of sunlight)

The method of course relies on the webcams not having an IR filter..

if you want to try this approach, the eyes should stand out when thresholding the image with the IMAQ Threshold, (do it on the luma color channel).
To get the coordinates use shape detect, or find pattern, or peak valley detector.

for face tracking, you can start with looking for the nose and mouth, as it typically will be seen as a vertical line, and a horizontal line below it, and must be placed somewhere in between and below the found eyes.

good luck

[Klemen]

Hello,

if anybody needs a starting point for face and eye tracking you can take a look at the following link:

https://forums.ni.com/t5/kl3m3n-s-blog/Real-time-face-and-eye-detection-in-LabView-using-OpenCV-Harr...

There is an Labview example code also attached. For further information regarding the algorithm see the OpenCV documentation.

Best regards,
K

[Klemen]

The page will be up shortly, since I made a small change (added two words!!!) and it needs to be re-moderated.

I really wonder when NI will find a better solution than this.

Best regards,

K

[alfie90]

Hey all!

Thanks for all the suggestions. I ended up using OpenCV for my application since it was easier to use and I was on familiar territory by using the C++ API.

I got it to work and aced the course!