Log in or register to post comments

Detecting objects in images other than camera stream

January 4, 2013 - 9:18am #1

Hi Vuforia Team,

firstly thanks for a great framework. I used the Vuforia sdk in my last project on iOS with great success, we had compared several similar toolkits, and vuforia seems to me to be the best by far.

I do have a question though. Is it somehow possible to track objects in imagery that does not stem from the live camera stream? The background of this question is, that I'm using opencv for tracking other things, such as faces and bodies. And I'd like to take the frames I receive from opencv and let them be analyzed by Vuforia at the same time, because in its field Vuforia outperforms opencv by far. So the ideal solution for me would be to be able to load a trackable set and manually submit new frames in some defined format to vuforia to then receive a list of transformation matrices of the found trackables.

Is this possible at the moment? And I so, could you point me in to the right direction? I've read through the code of "QCARutils.mm", but to me it seems that there is always some kind of camera connection involved.

Thank you for you time!

David 

Good.

January 9, 2013 - 2:46am #8

Good.

Thanks for your fast answer.

January 9, 2013 - 2:39am #7

Thanks for your fast answer. I'll give it a shot and let you know for future reference!

Glad that I was helpful.I

January 9, 2013 - 2:37am #6

Glad that I was helpful.

I don't know how to convert to CV:mat, but yes you can get the QCAR::Frame from the QCAR::State and then get a QCAR::Image from the QCAR::Frame; this is straightforward to do.

once you have the QCAR::Image, you can get the width, height and the pixel buffer (see API reference:

https://developer.vuforia.com/resources/api/class_q_c_a_r_1_1_image)

 

One thing to note is that the QCAR::Frame contains several images (not just one) in various formats and resolutions (some of them are for instance in RGB and some of them in grayscale; normally the first 2 or 3 images in the list of images for the Frame are high resolution, and the remaining ones are at lower resolutions).

I hope this helps.

 

 

Hey Allesandro, Thanks for

January 9, 2013 - 2:09am #5

Hey Allesandro,

Thanks for your answer, this really points me in the right direction. As far as I can see, I can grab the current frame (and thus the current image) object from the shared state object in QCAR. Do you have any resources that may point me in the right direction for converting a QCAR::Image into a cv::mat? That would be great. If not I'll figure it out somehow.

Thank you!

David

Hey Allesandro, Thanks for

January 9, 2013 - 2:09am #4

Hey Allesandro,

Thanks for your answer, this really points me in the right direction. As far as I can see, I can grab the current frame (and thus the current image) object from the shared state object in QCAR. Do you have any resources that may point me in the right direction for converting a QCAR::Image into a cv::mat? That would be great. If not I'll figure it out somehow.

Thank you!

David

Hi, Vuforia does not support

January 8, 2013 - 3:11pm #3

Hi, Vuforia does not support the processing of external images other than the camera video frames.

(actually it is possible to do the opposite, i.e. extracting the video frames using the QCAR API and feed them into OpenCV for instance).

If you find this desirable, you can post this in the wish-list (maybe for future versions):

https://developer.vuforia.com/forum/general-discussion/wish-list

 

I'll bump this, because for

January 8, 2013 - 2:12pm #2

I'll bump this, because for some reason this post went straight to page 2. Sorry about this.

Log in or register to post comments