Live data from Hacker News

China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

techcrunch.com

31–40 of 153 posts

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#31

Earlier quoted context omitted.

The University of Tokyo could afford to do that because another nation had taken over responsibility for their defense. Who's going to be that protector for the US?

Postulate: how does the US being able to more accurately track people via facial recognition make me meterialy safer as a US citizen? Because for me, that argument has a whiff of 'think of the children'. I don't think the NSA has done anything to make me safer, and I haven't seen any convincing proof to show they have done anything to help my safety. At this point, what is actually accomplished by the US military get…

Going to note that I am largely against facial recognition tracking to us citizens before I say this post.

Being able to track and find individuals, anywhere in the US where there is a camera, will significantly reduce crime. Cameras would cover a larger swath of territory, and we would be able to find potential terrorists, criminal suspects, missing people, human traffic victims, etc. far more effectively. We would likely be able to identify people acting like they were about to commit a violent crime too, but that’s just an assumption based off my experience.

Does large scale surveillance help with making individuals safer? I think the answer is obviously yes. I disagree with it because it has significant side effects, and doesn’t address the underlying problems CAUSING cultural violence and international crime.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#32
post #24

> For ordinary people we will only extract their data when they need our help. > When they don't need help, we won't gather their data and it remains only in our big database. In the same sentence: "We won't gather their data" and "The data that we gather remains in a database".

It's a common dialectic pattern in law enforcement circles world-wide that data at rest that is never touched is data that doesn't exist.

That angle is based on the idea that this data "may as well not exist" but it ignores the many issues with indiscriminate data collection: the data might leak, or management might change and invent undesirable "new" purposes for old data - which, within this theory, then begins to exist retroactively.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#33
This article is 2 years old, but quite relevant to current situation.

Deep learning and computer vision has progressed quite a bit in the past 2 years, mainly due to incremental algorithmic advances / improved engineering. Unlike fundamental advances, these can be scaled up with increased investments in a more predictable way.

Face++, the unicorn company mentioned in this article, is actually the product division owned by Megvii, a machine learning startup based in mainland China. They are planning to IPO soon (see: https://reut.rs/2L7zzGn ) to raise more funds from overseas investors (hence IPO'ing in Hong Kong).

There's some irony when a China-based computer vision startup working on facial recognition technology used in Xinjiang is filing for IPO in Hong Kong, rather than in mainland China.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#35
The interesting part is not the story itself, but that they chose to demo it to the BBC. Marketing at its best, they want everyone to be fully aware of their capabilities. Wouldn't be surprised if they were overstated, so as to keep citizens in check. The BBC willingly plays along and does not even question the capabilities either way.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#36
post #35

The interesting part is not the story itself, but that they chose to demo it to the BBC. Marketing at its best, they want everyone to be fully aware of their capabilities. Wouldn't be surprised if they were overstated, so as to keep citizens in check. The BBC willingly plays along and does not even question the capabilities either way.

You can bet they only demo things if they are 100% sure it will work (in a very controlled environment not unlike North Korean "tours"), and even then often it will fail because of Soviet-style bureaucratic incompetence.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#37
This tech works, and the only reason we don't have this in the US is because of our cultural and political aversion to mass surveillance.

I'm genuinely curious of what this leads to in China, because this tech is easy enough to deploy if you're willing and politically able. What I'm cynical about is that I don't think China is competent enough to safeguard the data it collects, or forever keep it out of the "wrong hands" like the last guy in the video claims.

It's hard enough for US government agencies to protect data that should be privileged (police often run illegitimate checks on license plates), I can't imagine China can prevent its various factions from abusing this power.

It'd be ironic, but the incompetencies of their bureaucracy could actually lead to their own downfall by way of this technology (their mass surveillance apparatus) somehow getting coopted by dissidents. There's a misconception that China somehow has an iron grip over its people because it's authoritarian, but if that were true they'd have weeded out corruption.

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#39
post #4

It's not just about this, but given all that is happening in the space of facial recognition, I think it would be really valuable for researchers to start looking at it like they would weapons research. Almost entirely nefarious applications. After WW2 the University of Tokyo banned weapons research from happening in its research labs. I think we can be drawing more lines on research like this and take a stand agains…

Facial recognition itself isn't the problem. Arresting good people based on bad laws is the problem.

It sure helps not to hand the tools directly to the oppressors.

Of course stopping all facial recognition doesn't suddenly make a totalitarian regime disappear. But having it makes some things way easier than it used to be.

For what? automatic tagging of faces in your photo library?

Re: China’s CCTV network took just 7 minutes to capture BBC reporter (2017)

#40

Earlier quoted context omitted.

The University of Tokyo could afford to do that because another nation had taken over responsibility for their defense. Who's going to be that protector for the US?

Postulate: how does the US being able to more accurately track people via facial recognition make me meterialy safer as a US citizen? Because for me, that argument has a whiff of 'think of the children'. I don't think the NSA has done anything to make me safer, and I haven't seen any convincing proof to show they have done anything to help my safety. At this point, what is actually accomplished by the US military get…

> Postulate: how does the US being able to more accurately track people via facial recognition make me materialy safer as a US citizen?

I don't think tracking people via facial recognition will significantly help, but some similar technologies might. Recognizing camouflaged tanks (or disturbed ground concealing mines, or flares vs engine exhaust, or...) would, though. And I don't think those techs are separable, if that makes any sense.

> At this point, what is actually accomplished by the US military getting even fancier ways of killing people? From how things are shaping up, even if things come to blows b/w US & China it won't be a shooting war. It's looking like it will be similar to the cold war: neither adversary can dare attack the other, so we end up with a series of proxy wars and nitpicking at the adversary.

And if you think that neither country will be able to afford to attack the other and thus stop preparing to be attacked... Well, that feels like it'll invite an attack. Not to mention that some of those proxy wars might end up being "Korea mk2" or "Taiwan", both of which would make conventional military expertise extremely valuable.

Post reply on HN