Log in or register to post comments

Enhanced capabilities

November 10, 2017 - 12:45am #1

Hi Vuforia !

I have started developping some recognition apps for a few weeks for the hololens, trying the image and object recognition and I'm curious about how these tools will get even better. Let me explain ;)

The recognition works very well with images like QR codes but is quickly very limited when the image has fewer contrasts or when we try to recognize a specific form but with round shapes, or when there is light reflection.

The object scanner tutorial shows great results with the toy car in the video, but is unefficient with objects like bones that have specific shape but no contrasts at all... I first thought it would create a 3D volume and thus make the hololens able to recognize anything, then I read it was not that easy...


My question is : where is the limitation for now ? Do you wait for technology to improve? (depth camera on smartphones?) Or do you just need time to develop your tools?

And also, can we imagine one day a hololens scanner to create 3D recognizable objects?


Thanks for reading !


Enhanced capabilities

November 10, 2017 - 8:54am #2

Hello JP,

The contrast and light reflection is a limitation of the feature as Vuforia is looking for feature points on the target to detect and track. Depending on the amount of reflection and the number of feature points on the target, it is possible to still get a decent tracking experience.

In terms of your use case of tracking an object such as a bone, that might be a good use of our upcoming Model Targets feature. This feature detects objects based off of the geometry of CAD/3D model data rather than texture-based feature points on the model. Model Targets are currently planned for Vuforia 7 (https://developer.vuforia.com/), but you can apply for our Early Access Program (EAP) if you are interested in using this feature before the commercial release: http://info.vuforia.com/modeltargets


-Vuforia Support

Log in or register to post comments