List
Improve Image Trackable Accuracy
Posted Date: 2018-08-24 6:27     Edited Date: 2018-08-28 0:54     Writer: inactive

Hi Team,

We are using MaxST’s Image Trackable feature in our project. We wanted to differentiate between two appliances (Images Attached) which have a lot of similar features through Image Trackables. The detection happens accurately around 70% of the time. Can we know what exactly is causing this confusion and what are some ways through which we can improve detection accuracy?

Thanks

Posted Date: 2018-08-27 1:37     Edited Date: 2018-08-27 1:37     Writer: kscho

Hello.

Thanks for your interest in MAXST SDK.

Unfortunately, two targets are too similar to recognize exactly.

Sorry that it is hard to suggest any tip for improvement.

Thanks.

 

John,

MAXST Support Team

Posted Date: 2018-08-27 4:37     Edited Date: 2018-08-27 4:37     Writer: inactive

Hi John,

Thanks for your response. Will the detection accuracy improve if any pattern/sticker is attached externally onto one of the appliances to make it more distinguishable from the other?

Thanks,

Aiyappa

Posted Date: 2018-08-28 0:54     Edited Date: 2018-08-28 0:54     Writer: kscho

Hello, Aiyappa.

Attaching any pattern/sticker shows a better result, but it seems that it is not perfect.

If a feature set is a subset of another feature set, it is possible that mis-recognition can occur.

Thanks.

 

John,

MAXST Support Team