We recently bought a cloud plan license for one of our projects and are experiencing some problems with the image tracking.
Our project consists of an app that uses Vuforia features to recognise the pages of school books in order to give access to multimedia content. The books are available in two versions (one for teachers, one for students), with little differences between the pages (you can find examples in the attachment files).
At first we only uploaded the teacher version which had the better rating, due to the higher number of graphic elements in the page. However, in time, we found out that the students were having trouble recognising the pages with their version. To solve this problem we uploaded the student version as well, hoping it would solve the problem. It didn't solve it, made it worst. Now the two targets are conflicting, but only in some devices (with less specs), or in poor light conditions.
Do you have any suggestions on how to overcome this issue? How does Vuforia deal with a case like this where there may be conflicting trackers? Does it respond with the best match, or with the first match?
The output for teacher and student is exactly the same, so we are only hoping to find the better solution in order to provide a good tracking for both types of users.
Many thanks for your patience and cooperation.
OpenGL ES 3.x means you’ll have access to a broader range of textures and shaders that you can use to create high-end effects in your apps.You can also take advantage of rendering improvements and power optimizations made for recent mobile devices.Cloud Recognition users will benefit from updates designed to improve recognition speed and accuracy.We are proud to release the latest advancement in the world's leading augmented reality platform.Custom Writing Service for Students – order online best custom paper.Order now your's discount coupon here https://beforewriting.com.
Thank you for your detailed description and supporting images. Reviewing them, I can easily see how our detector is getting confused as there are few visual differences between the two examples.
For either on-device or cloud database, our detection algorithm will extract features from an image and use a database-type lookup to match them with a known reference (i.e. the images that you want to detect, and are uploaded via Vuforia's web services). The features extracted create a digital "fingerprint" of the image. As the "fingerprints" between two images become more similar, the probability for an exact match lowers until it becomes even between the two. I believe this is the issue you're facing.
In the short term, my only suggestion for resolving the detection confusion is to place the student and teacher edition pages into separate cloud databases.
In the medium to long term, I would suggest that the publishing company consider VuMarks. These would eliminate detection confusion, and would be a call to action for users such that they know which pages are linked to additional digital content. Lots of potential here.
Are you sure you want to delete this message?
Are you sure you want to delete this conversation?