Log in or register to post comments

Testing if Markers too similar to each other in dataset

July 17, 2018 - 4:30am #4

Hi there,



I have a dataset with just over 62 image targets. All of which get 5 stars.

However, a lot of them share the same feature points. Therefore when testing on device one marker can get confused for another.

Apart from manually testing, is there an easy way to determine that markers will clash?

Thanks

Testing if Markers too similar to each other in dataset

July 17, 2018 - 1:40pm #3

Hello,

Unfortunately, we don't provide a tool for this type of analysis.

Thinking out-of-the-box, you could utilize our VWS APIs to perform a similar analysis: https://library.vuforia.com/articles/Solution/How-To-Use-the-Vuforia-Web-Services-API.html#How-To-Check-for-Duplicate-Targets

Just an idea...

Thanks,

Vuforia Engine Support

Testing if Markers too similar to each other in dataset

July 18, 2018 - 2:19am #2

Thanks very much for the reply, I will take a look.



Along similar lines, is there away to get vuforia to only trigger recognising a marker if closer to 100% of the feature points are infact found.

At the moment Vuforia is super clever and you can hide more than 50% of a marker and it will still recognise it. I unfortunately need a more dumbed down vuforia that will only recognise a marker if say 90% of it is found.



Is that possible?



Thanks

Testing if Markers too similar to each other in dataset

July 18, 2018 - 8:17am #1

Hello,

Unfortunately there is no way to tune Vuforia's detection parameters in the way you've asked.

Our computer vision algorithms have been engineered to maximize performance and robustness.

Thanks,

Vuforia Engine Support

Log in or register to post comments