Hacker News Reader

Comment by sandworm101 | Hacker News Reader

What matters more than the number of cameras is how/if the cameras are integrated. I don't much care that the local gas station has cameras tied to a 30-day DVR. I expect that. I don't expect them to be connected to facial-recognition AI, networked to the national police, and be feeding my movements into a citizenship score system that will impact my future job prospects.

View on HN

Well, Atlanta has integrated over 10,000 cameras and 200 pole mounted automated license plate readers. The per capita statistics are distorted by the relatively small boundary of the City of Atlanta, but there are still a lot of cameras.

[https://atlantapolicefoundation.org/programs/operation-shiel...]

[https://www.muckrock.com/foi/atlanta-325/2018-vigilant-data-...]

While not networked to the national police I expect that many commercial security cameras do have facial recognition to identify suspected thieves and VIPs along with tracking movement habits through the store which is shared with advertisers and product manufactures to justify cost for higher profile shelf space.

>> tracking movement habits through the store

My understanding is that was a passing fad a few years ago. It was hyped by some startups but commercial users found it less than helpful compared to traditional metrics. I met a grocery store consultant (a real job) at a security conference. He said they used cameras to deal with traffic issues (even load between aisles) and predict cashier/customer ratios. They had abandoned the face/eye-tracking tech. Eye-level shelf space is better. They don't need an algorithm to keep telling them that.

They don't need CV for this. They can use beacons embedded in smart devices to tell which part of the building you're in. Worked with one of these guys.

They need cameras for theft. Tossing a person detecting algorithm on top is basically free (especially if you don't need it to do fancy things like track faces).

How would one calculate that score?

Investigative journalism.

Parked too far from pump -1. Didn't check air pressure in tires -1. Drove too quickly -1. Yawned too much, driving while tired -10 points. Did have correct national flag in rear window +1. That is all easily within current AI abilities.

You can ask Alibaba https://en.wikipedia.org/wiki/Zhima_Credit

Reply on news.ycombinator.com