> The accuracy of the results don't seem that great. For example, looking at the pictures on the wall in their sample, or the beams in the ceiling.

Do you mean the accuracy of the classification or the precision of the lidar scans?

In my experience the lidar precision on the iPhones is decent but not great, so the texture mapping can look a bit off at times.

I'd love to have these bounding boxes on my scans though.

I mean the accuracy with which it's locating the bounds. What is extra curious is it obviously supports rotated cubes, yet it often doesn't use them when it should, leading to overstating the bounds, as if it's over enthusiastically trying to put things aligned to some inferred axis.

This is obviously an attempt at the general case to apply cubes to anything, but what is disappointing is the performance on boxy objects is lower than I've seen on private NNs used for AR and CV for years (ironically enough on iPads), using just rgb and no depth.

I half think the exercise here was to establish if transformers were the way to go for this, and on the strength of that the answer would be probably not.