Image tracker - reduce false positives

I’m testing zappar on node.js with my laptop’s webcam, and sometimes the image tracker triggers when nothing even remotely similar to the trained image is in view. Is there any way to reduce / completely avoid false positives?

Hey @mmagarik,

It’s hard to tell what’s causing this without seeing your target image. I would suggest having a read over our what makes a good target image article :slight_smile:

Thanks for the tip!
I am working on improving the images used for training, but was wondering if there’s anywhere a sensitivity setting that would in general make the system more or less trigger happy, since in my current project there are all sorts of factors that limit how good the images can be.