Start a new topic

Tracking multi-image targets

Hi, 


We're trying to implement tracking of Merge cube - cube with 6 different textures on the sides, using Wikitude's native SDK. Currently we use ImageTrackerConfiguration with six separate images and compute cube transformation based on the tracked side. With this approach, tracking quality is good when one side is clearly visible, but tracking is easily lost when cube is rotated from one side to another, even by slow rotation. 

Are we possibly missing something in the tracker configuration, e.g. some parameters worth tweaking? Or maybe there is a better approach for tracking such markers available in the SDK? 


Best regards, 

Kirill


Hi Kirill,


Did you already try to work with our Object Recognition? Your use case sounds similar to what we show in our latest demo for our Expert Edition launch - https://www.youtube.com/watch?v=iak9zBhhoR4&feature=emb_title - and it sounds like Object Recognition might be the suitable approach. 


In case you work with Unity, we definitely recommend that you check out also our Expert Edition. Details on the features of this recent launched version of the SDK can be found here: https://www.wikitude.com/blog-augmented-reality-for-experts-introducing-wikitude-sdk-9/. But also our Professional Edition of the SDK offers a comprehensive Object Recognition and Tracking feature.


So give it a try (the technical documentation gives you guidelines on how to create an Object Target) and let us know if you have further questions.


Greetings

Nicola



Hi Nicola, 


Thank you for your suggestion, tried the proposed approach with object target. 

Unfortunately, for now it worked much worse in our case than tracking side images separately. Attached it the best point cloud we were able to get (from ~30 images). The main problem is that it doesn't seem possible to take images from all directions, including bottom side of the cube. And even without bottom side, tracking is established only when looking at the object from nearly the same angles as images were taken. 


Best regards, 
Kirill

Hi,


Could you please also share sample images of the object from the different sides? Then we can check suitability internally.


Thx and greetings

Nicola

Sure, here are two series of images with different lighting conditions

https://drive.google.com/open?id=1uNYXZKI2HhQnX72vGJLjlHNJNxvwRpZN

[Upd] First link was wrong, corrected link: https://drive.google.com/open?id=13AjnLv0ARaX-D-6ieEvQXVzp2NZxYA44

Hi Kirill,


We checked the images internally and our colleagues from the computer vision team mentioned that the object itself should be suitable for object reco. Here are some tips on how to create and improve the object map:


Extending the map with new images is always a good appraoch to enhanve the map and add areas to the existing one; This step-by-step approach allows you to improve the map gradually and test it accordingly. Did you try to extend the map with images of the missing faces?


  • The ideal image perspective in the case of the cube, would be to have images, that show 3 sides / faces (Having images just showing 1 side makes it harder for the algorithm to make the connection to which side it belongs and which sides are next to each other.)
  • Make sure that the images are sharp and all sides well lit
  • Create the map with some images from a specific angle, then go further around the object and extend the map gradually with the images from the next angle and so forth (if you have always 3 sides visible then turning the cube around to also have the 'bottom side' should work without problems)

Please try the above and let us know if the map improved.


Thx and greetings

Nicola

Hi Nicola, 


Trying the proposed approach with building initial map based on 3 adjacent sides and then extending it. 

First step works fine, an accurate map with three faces is created. 

But when extending it with series of images of another 3 sides (with two overlapping with previous sides, and one new), new side is not added and the map is only slightly rotated.

Tried with three different series of images, but they give the same results: https://drive.google.com/open?id=13AjnLv0ARaX-D-6ieEvQXVzp2NZxYA44 (folders 1, 2, 3 each contain series of photos of three adjacent sides)


Best regards,

Kirill




Hello, I have the same problem; any update about it?

Hi,


Could you please provide further details on your use case and the images you wish to track? 


Thx and greetings

Nicola

Hi,


I've already tried every method, i'm trying to track the six faces of a cube but all the times i have the same result, the software doesn't track the bottom face. I took like 300 pictures and i'm using a light box so that i'm pretty sure the images quality is very good and well contrasted. Here are some pictures as reference.


Thanks

IMG7647.jpeg
(1.29 MB)
IMG7631.jpeg
(1.35 MB)
IMG7618.jpeg
(1.33 MB)

Hi,


We're currently looking into it, on how the object target can be improved. So all sides except for one are recognized and tracked fine? For the side that is not tracked well - did you make sure to also have 3 sides well visible in the pictures you take and did you follow the tips we gave in this thread with extending the map?


Thx and greetings

Nicola

Hi,


I've already tried to extend the map many times and as i mentioned in the previous comment i took the pictures in the best way possible and from many different angles with 3 faces in view, now the 5 faces which are working have a lot of tracking points but the main problem is that the tracking is not stable when i go from a face to the other and i lose the 3d object all the time. Is there a way to improve the stability? Or should i take a different approach?


Thanks

Login or Signup to post a comment