We are trying to identify a target which is a shoe and attach a holographic marker to same. But we are seeing challenges with respect to consistently and steadily viewing the hologram on the physical object. There seems to be variations due to lighting condition, distance and angle of viewing from the Microsoft HoloLens device which is being used to view and identify the target (Shoe).
The exact image(left & right side of the shoe) has been obtained from the product website and uploaded to Target Manager (.jpg file <2 MB size). When the image of the shoe is shown on a phone screen, the holographic marker is seen on the picture of the shoe but again only at a certain distance and angle in a well lit room with fluorescent white light.
Any suggestions on how to improve the performance of the target tracking and recognition and integration with the Hololens is appreciated.
Should we be using a image target or an object target ?
Thanks
Vijay K
| Attachment | Size |
|---|---|
 71D+IyuN75L._UL1500_.jpg | 189.6 KB |
Hello,
To clarify, you are using a 2D, planar image of a shoe as a target and are finding limitations in the distance and angle at which it can be detected?
All computer vision algorithms, including those running inside Vuforia, are influenced by environmental factors such as lighting, motion blur, distance to camera, etc. Additionally, the quality of a target, like those recognized by Vuforia, can be degradated by specular reflections - when a lighting source, reflecting off the surface of a target, introduces an occlusion of the features used to detect and track.
My suggestion is that during your testing, pay close attention to environmental factors that could be affecting performance.
Thanks.