• Any Pronons Are Fine!

30-something
Black, andLikes Comics and junk.


nys
@nys

not to defend amzn but technically they were mostly reviewing footage after the fact and correcting mistakes the system made. the system was still responsible for your checkout.

all of this is for the same reason they bought whole foods and immediately installed similar camera systems. they were building a corpus of video tagged with the correct result so they could train/test their models. at whole foods they used check out as the source of truth and with amazon go/fresh they used manual tagging. in the algo world well tagged data is worth its weight in gold (how much do bits weigh). the still still absolutely bonkers part is that they tried for six years.

what i’ve heard through the grapevine was that the biggest issue was that the compute cost required to accurately—in realtime—track individuals picking up all sorts of objects (even tagged with rfid) was nigh impossible to bring down enough to be worth it or scalable. i had an amzn employee drunkenly mention that at one point each checkout required over 10 minutes of gpu compute per minute of an individual shopping. it might’ve been even higher—i want to say 20, but that seems too ridiculous even for amzn.

add on top of that the fact that they tuned it to really avoid false charges so it consistently didn’t charge for items and blamo, huge money sink.

no wonder they failed to sell it to other stores:

  • every customer in your store costs you money
  • it misses a bunch
  • it was silly easy to trick
  • inventory is also a nightmare
  • risk management probably just said “no chance”

You must log in to comment.

in reply to @stu's post:

I was thinking about this and I think part of the reason is that self-checkout already exists in stores, so it wouldn't have gotten them press coverage about how "weird" it is to shop there if you had to stop and tap your bag/items on an NFC tag reader, but I bet another part is that people steal from self-checkout all the time. If you know how the system works, you can figure out how to get around it, but a "magic" system that just knows what you're carrying out of the store and you don't know how, then you can't figure out how to trick it.

i'd never been to one of these places so i always just assumed it was like, scanning things as you go along and getting it charged to your card when you leave or something. the reality of it being some guys in india watching you on cctv and tallying it up by hand is such a uniquely dystopian and stone age solution to the tech magic sales pitch we got you can't help but laugh at just how absurdly bleak it is

yeah i also assumed it was something like "self-scanning as you go", which according to the article is apparently the backup plan for stores who want to use something like this in the future. Insane that this wasn't the idea from the get-go.

"Just Walk Out relied on more than 1,000 people in India watching and labeling videos to ensure accurate checkouts. The cashiers were simply moved off-site, and they watched you as you shopped."

Oh my god just end me

in reply to @nys's post:

if the 10 minutes of GPU compute for 1 minute of individual shopping is close to accurate, I wonder how many years of raw compute improvements will be needed before this might actually become doable economically.

from my understanding it’s basically infeasible without a novel approach to the solution. also that comment was 2ish years ago so i assume it got better