at the beginning I would like to say that VuMark is nice technology that works pretty well. But still I feel like I could achieve better results if I get better control over recognition algorithm.
My use case is similliar as with lego nexo knight shields. I've printed vumarks on cards and only thing I need is to recognize them. I do not need to track them, only recognize. I didn't find way how to turn of tracking but it's not really problem. I just don't use any tracking info I could get. First problem was that I need to use front camera but I solved it by using custom image source. Front camera has worse resolution than back camera but I printed vumarks in adequate size with quite big bits and good contrast, so even if the card is not well focused, the Vumark is recognized.
But what I need is to recognize VuMark as fast as possible. Result is not bad but still even if the Vumark is well focused, there is a little delay before the VuMark is recognizied. And it feels impossible to recognize Vumark during movement (even slow one). The card have to stay still for little while to be able to recognize Vumark.
It feels like the recognition algorithm wants to be really sure with the result and maybe checks the Vumark few frames in the row before it decides that Vumark is really seen. Is there any way how to tweak this behavior? Or is maybe possible to get access to some extra API or event source code with Pro licence?
Thanks for the answer