Design: Time-Lapse inventory system
Having moved a few times, and done a few things in our home, the need to keep track of things has periodically surfaced.
Goals
Easy. It can't be a lot of extra work, both because the effort is unlikely to be rewarded to profit, and because if it does, that boon is for future you and in any case we're exhausted.
Unobtrusive. It can't require interruption of work to document, because constant interruption is an extremely bad way to get work done and any effort to document is undone in a moment when you discover that that box fits perfectly over there instead.
Available. It shouldn't require too much in the way of prep or props, because we're busy moving stuff and doing all the things we normally do in our lives. And it definitely can't require stuff ordered online that's stuck in the package undelivery en-/re-/mis- route circus.
Design
Step 1: Prep
Some minimal prep-work goes a long way, particularly if we'd be using identical moving boxes. We can print QR barcodes with (large) random numbers or words on them, in a narrow enough format the they can cut out easily to the point of being sloppy, and still be just sticky taped over. If appropriate, these should perhaps be made in a few copies each, or just near each other, so they can be stuck to related things, like different sides of the same box, or the inside and outside of the flap, or whatever fits. (Pick a random number in the billions, and use it, and the three numbers after it. They're distinct, but also clearly in a group.)
You would need a phone or computer with a good enough camera. This turns out to be a bit trickier than it first appears, because we really would want resolution and quality good enough to read those barcodes right out of the pictures, and webcams are impressively bad.
Step 2: Capture
Set up a phone to capture a time-lapse of the packing.
If this gets realised as an app, capture video and select which frames to use, or if we have the space, just leave it filming, and run the same frame selection algorithm after the fact. Or just run a time-lapse photo app, you're likely to have more than enough photos anyway. If you have those barcodes, show the barcode as you're packing a box, then stick it on the outside, or whatever works. (For example, related codes on the inside flap of a moving box, visible when packing it, and on the outside of the box, when placing / stacking the box itself.
Step 1: Winnow and refine
Given a really large and really dull photo album, the next step is try to get rid of as many unhelpful photos as possible to improve the signal-to-noise ratio. Some can probably be discarded by mechanical means, such as detecting duplication, excessive motion blur, and similarly obvious issues. A fun side note here, is that this is something that can likely be improved upon in the future, in a way that benefits earlier captures.
In this step we'd also want to do things like detect barcodes in the pictures, not just the ones we added but also on, for example, paperback books. A surprising number of our possessions have them, and if we can read them, why not? AI Object detection is also a possibility, but by no means a requirement.
Step 2: Use
In a fully matured system, you could just ask out loud, "Oh, Omnipresent Eavesdropper of the Abhuman Corporation, where is my red stapler?" and it would tell you, "It's in the attic, behind 17 other boxes, 8 of which are overloaded, and two are wedged in very tightly... So it's basically thrown away, but with extra steps". Closer than this (disturbingly near) science fiction however, in a first version, we could just scrub through the time lapse to find what we're looking for.