I’ve written a few times on this blog about our library’s augmented reality (AR) game, which we’ve been using during information literacy sessions. The game has been mostly successful, but we’ve had trouble with the computer vision part of it: sometimes the game identifies markers that are not actually there, and therefore shows shapes in the virtual space at the wrong times (we’ve been calling these false positives). Anyhow this was very vexing, and we had put a lot of effort into fixing it without much progress.

But this week I had a big breakthrough. I swapped in a new marker recognition library called Mind AR. It is way better than AR.js, by leaps and bounds. Here are some of the advantages I’ve seen so far:

• No false positives.
• It is possible to use much more detailed markers.
• Far less CSS hackery is needed to get the games’ menus to display properly in the virtual space.
• It is a drop-in replacement for AR.js, so our A-Frame code can remain pretty much the same.

Resolving the false-positives problem is significant because it allows us to move forward with some of our goals: we can build out quiz questions and other such dialogs, which would not have been suitable at all if they had been appearing due to false positives.

We’ve also moved the game from LibGuides to PythonAnywhere, which has opened up some additional possibilities. All in all a good week for the AR game!

This entry is licensed under a Creative Commons Attribution 4.0 International license.