NYPD can use more than 15,000 cameras to track people using facial recognition | Hacker News Reader

NYPD can use more than 15,000 cameras to track people using facial recognition

amnesty.org

138 pointsraybb88 comments

I am so glad many cities and even Cantons in Switzerland have decided that such a thing should not be permitted. [1] Such a distopian future is not something people want anywhere.

I assume it is still legal in NYC to wear clothing or head gear that would make face recognition difficult if not impossible?

[1] https://www.gesichtserkennung-stoppen.ch/news/grosse-erfolge

The normalization of face-masks in public have been amazing for aiding anonymity.

too bad one can be identified also from the way they walks [0]

[0]https://www.cnet.com/science/china-surveillance-tech-ids-peo...

I don't think there's a nationwide database of walking gate fingerprints facilitated by your local DMVs.

Not if they apply https://github.com/topics/gait-analysis &

https://github.com/topics/gait-recognition

https://securitytoday.com/articles/2018/11/12/new-surveillan...

https://recfaces.com/articles/what-is-gait-recognition

If they (can) do so ATM (at large scale and in real time) is another matter, but only a question of time, IMO.

Not sure how effective it would be in reality, but in Cory Doctorow's Little Brother they work around gait analysis by putting some pebbles in their shoes.

As a temporary means of distraction, maybe. But it wouldn't really change the https://en.wikipedia.org/wiki/Kinematics, ratio of joint to limb length, and so on. You'd just sway a little bit different, but not that much.

Then there is https://en.wikipedia.org/wiki/Wide-area_motion_imagery AKA https://en.wikipedia.org/wiki/Gorgon_Stare , https://en.wikipedia.org/wiki/ARGUS-IS , the digital signature of your gadgets in the EM-spectrum, where their sudden 'disapperance' might raise flags, leading to 'replay' of the region for further analysis.

But there is always hope in rising incompetence of ever growing institutions and organisations, hollowed out from the insides by the parasites they attract.

Yeah a lot of people think funding innovative new walking styles is why they exist, but they ministry of silly walks has other more sinister purposes as well.

https://youtu.be/eCLp7zodUiI

Crime is one of the most important voting issues. Republicans had a resurgence in NYC because of it.

Alternatively one could say Democrats lost ground for being too soft on it.

Seriously though, your comment makes it sound like Republicans benefit from high crime. No one benefits from high crime.

Or rather they weren't soft enough. Republicans are just cruel and unjust because that usually benefits those in the in-group. Democrats compromise so much they are left with the worst of both worlds.

They couldn't decriminalize drugs enough or enable world class social welfare programs when they did get majority control over legislature. But voters are ultimately to blame, they want their cake and to it too.

> But voters are ultimately to blame, they want their cake and to it too.

As a voter, you're given the choice between terrible and truly awful. No matter which you choose, it's your fault?

Frankly, when I buy a new kind of underarm deodorant and later I stink, I feel I'm more to blame than I am for my political vote and any lack-of-outcome. And I'm inundated with deodorant choices - it's harder to pick.

I remember when New York wanted to lynch the Central Park five. Reason and idealism seems to vanish when people feel threatened.

People engaging in petty "affront to the authority of the state" types of crimes benefit because in an environment where you have plenty of real crime to go around there's no need for cops to initiate the kind of police interactions that end up with Eric Garner getting choked out or a bullet in a man who has a bench warrant for failure to pay some BS.

I don’t think republicans want this. I think all they want is the criminals already identified or caught to actually be prosecuted and not released back on the streets.

They are politicians, they want to look tough on crime no matter the cost. They make the laws. War on drugs is their creation, half the criminals are created by them. Rudi got popular because stop-and-frisk in NYC, for no legitimate reason cops would stop random people and frisk them. This social classist hostility where crime is what the fund-raising class it is and the solution is hostility against "those people" who are the least likely to be in their party is their MO.

Republican voters want whatever “get tough on crime” aligns with. “Repeal bail reform Derp Derp”

The “freedom” stuff is bullshit and narrowly scoped.

I didn't check if this research is still pursued or abandoned due to the rise of face recognition and other metric learning/re-identification technologies, but a few years ago a number of faculty at ETH Zurich were quite heavily involved in researching CCTV-based person tracking, which would work even against people taking measures against face or gait recognition algorithms if they weren't able to leave the CCTV-covered area. It was insinuated that various external state agencies, not all Switzerland-based, were specifically funding this research.

So, while the citizens might not want it, faculty seem perfectly happy to keep developing such technology for other clients.

Link to the research?

Some of the theoretical work is available with keywords like 'multi-object tracking' or 'multi-object detection' and also applied to tasks such as pedestrian tracking for autonomous vehicles, but there was also a bunch of applied work building and evaluating such CCTV systems which I can't find any links to now, sorry.

Just wear a cap, shades, and a covid mask.

Illegal + doesn't work against modern solutions incorporating e.g. gait analysis.

Great. And they still can't catch that guy who robbed the bodega on the corner, and people casually ride the subway without paying every day.

NYC's latest shooter turned himself in, the cops couldn't even handle that: https://abcnews.go.com/US/nyc-police-search-gunman-brooklyn-...

Public transit should be free

In many cities I think that this is a good idea, but in New York I worry about how effective this would be because of the outsized number of non-resident, non-tax-paying riders (tourists) who use the system.

Most of the tourists will be paying a nightly hotel tax (which could be adjusted).

I'm completely fine subsidizing tourists' use of the system. Every subway ride is one less rideshare ride; less rideshare rides mean more space for buses, bikes, and people walking. Well worth the tax paid!

That farebox revenue needs to cover operational costs for public infrastructure seems just crazy to me. The fire department doesn't need to make a profit on putting out fires. Public parks don't need to make money by charging an admission fee. It's just something we pay for for public good. I'm pretty sure that public transportation fares are something nobody has ever really thought about; the networks used to be run by private companies, and when they were taken over by the government (which put them out of businesses by building highways) nobody ever thought to get rid of the fares. It's certainly time to think about it.

Tourist tax. Like when you check-in to your hotel they ask you to pay 15 euro.

This already exists, and there are various excise taxes the city levels on things tourists are likely to consume too.

Interestingly there has been a parallel conversation with EVs since money for road maintenance typically comes through gasoline taxes. The idea being that the more you use the more you pay (more driving == more gas, heavier vehicles == lower mpg, etc).

Wouldn't you want tourists to travel as efficiently as possible so they leave lots of money at various places (which the city participates in via sales tax)

I'd want locals to travel as efficiently as possible too- to get to their jobs.

> non-tax-paying riders (tourists)

They're literally carrying money into the local economy, yet one should worry about them getting a free ride on the subway along with everyone else?

A taxpayer may need months to generate the same amount of wealth a tourist leaves behind in a single week.

Making public transport free/cheap for only tourists would make more sense. In fact some places do just that (usually bundled with discounts/tickets to popular attractions).

Tourists pay sales tax, Uber tax hotel tax etc

Why and how?

Edit:

1.What will be the second and higher order consequences? See the student loan crisis and ballooning costs of higher Ed due to admin bloat.

2. Is it fair in a society with more free basics to require more responsibility from the public?

3. It is easy and popular to mandate feel good stuff than reason about that mandate.

The benefits of subsidizing public transit are obvious: less congested roads, more economic activity, fewer cars polluting the city.

Free public transit is just the same idea taken to the extreme.

> 2. Is it fair in a society with more free basics to require more responsibility from the public?

I was curious what you were thinking with your second question above, in the context of free public transit. I wondered if the need for responsibility would be a wash when comparing transit vs driving, affecting both?

For driving you need to pay taxes, pollution tests, fees, buy insurance, vision tests, and take a written test and a driving test. And continuously update your license and registration. All this is before you can drive. And there are road rules too.

If we have free transport, can we at least require minimum rules for safety?

E.g., no open drug use and harassing others.

Rely on holy spirits for safety is not an answer.

https://www.nytimes.com/2022/11/04/nyregion/new-york-subway-...

It's nice sentiment, but studies have found that making transit free just tends to cause existing transit users to ride transit more frequently for shorter distances (instead of walking) but doesn't tend to encourage new users of transit.

What time period did they do the studies on? Suppose I drive a car because it has some advantages over paid public transit. Now when they make public transit free, I still have my car. I'm not likely to switch within a short time period. When it comes to moving home, getting a new job or a new car then I may become a new user of transit.

You raise a valid point; this lack of long term effects is also my problem with most basic income studies. That said, if you're making a decision to sell your car and move to a place near a transit stop, I'd argue that the $2.50 or whatever a train/bus ticket costs isn't going to be a primary factor in your decision.

Consider Japan; everyone uses transit there, and train tickets are generally quite expensive. But, people pay for it because:

1) Almost all development is high density near transit stops

2) The trains are fast and frequent

If we want people to use transit, low fares are a definite nice to have, but the real key is we need to stop with horrible car-centric zoning policies and low-density construction.

It makes transit more accessible to people without money.

And it saves a good chunk of money on fare collection, processing, and enforcement.

Thankfully, since the beginning of the pandemic, most mass transit in the U.S. is de-facto payment-optional.

Studies paid for by who? The car lobby

…okay? That seems fine?

The problem then being that it becomes so over used and underfunded it collapses...

When free things are oversubscribed, either a queue or a scrum forms.

Which version of free are you talking about? Don't pay to maintain the equipment and enslave the workers, "point a bunch of guns at wealthy people and ask them to give us nice things?", ad-supported, donation-based (National public radio on wheels/rails)?

FTFA: "Data on commercial cameras available at the time suggest that it is likely to be a Pan, Tilt, and Zoom (PTZ) model containing a 6–134 mm varifocal lens able to film at 4K or 8 megapixel resolution."

This is so wildly inaccurate that it really makes this whole thing seem highly suspicious.

The average CCTV camera in a municipal deployment, especially ones from several years ago, would be a fixed 1080p camera with an adjustable 4-8mm lens, which is set at a fixed focal point upon installation. In other words, not a PTZ, not a long range lens. The mounting height is usually 10+ feet high, which overall makes it really poor for face recognition purposes.

This is not to say that many of these cameras cannot capture at least some faces with a level of detail that is sufficient for face-rec, but this camera network is nowhere near the ability to track a person/their movements throughout the city.

Based on being on an NYC grand jury where we had to watch a fair amount of NYPD submitted surveillance footage:

- NYPD uses PTZ, and the mount points aren't always fixed. They're also on poles which are connected to a mobile platform that is then deployed onto sidewalks / in parks etc.

- Ability to track across multiple cameras exists today (this is intended to address density of cameras, not how tracking is done). However, there was no clear indication how automated this was. We were simply shown "Suspect on camera 1, and then you see them leave frame, and here they are on camera 2"

- Faces could be seen fairly clearly even at night in a decently lit park setting.

My gut feel was that as the hardware improves, 100% automated tracking of everyone is only a question of when.

New Haven, the city where Yale is and a much poorer city than NYC had PTZ surveillance cameras deployed when I lived there close to a decade ago.

Cities can afford PTZ cameras and the backends to store mountains of high quality footage, the average small town municipality can't

There will almost always be some PTZs, but they are more of the exception than the rule. Also, when they are PTZs, they will tend to have smaller imagers and be 1080p max in previous eras (4K is more common in the last couple of years). The assumption that article makes about these cameras all likely to be 4K or 8K PTZs with the longest lens options is just absolutely wrong.

> our modelling suggests have the capacity to track faces from as far as 200 metres away (or up to 2 blocks).

Does anyone know how true this is? I watched the video and it's pretty convincing but compared to the security footage so often seen in the news this seems like it would be insanely high resolution.

I thought such footage is bad quality because nobody wants to store months of 4k footage of tarmac. Maybe they process it immediately for facial recognition, then store a lower resolution version? Or they just use AI "enhancement" and pretend they got the right guy.

Or just use AI to detect faces (without recognizing them), and store the high resolution faces for later processing.

They don't really care who citizen <random alphanumeric string> is until they get uppity at which point they'll figure out who's name to put on the arrest warrant. Cheaper to process now and store the row in the DB than store the whole images for years.

See my other comment, it's basically not true at all, IMO.

NYC has several thousand LinkNYC kiosks run by Google's Sidewalk Labs, located on major streets, https://news.ycombinator.com/item?id=18024123 & https://cryptome.org/2016/06/linknyc-spy-kiosks-installation...

> Each kiosk has three cameras, 30 sensors, and heightened sight lines for viewing above crowds ... According to privacy watchdogs, the rollout of the kiosk’s cameras have shown how the mission has already expanded beyond its initial purview. In 2016, LinkNYC disclosed that the kiosks “may” contain cameras; by 2017, the cameras were operational. LinkNYC’s privacy policy states that cameras do not keep video records for more than seven days and that the camera footage is used to “improve the services.” But opting out is not an option: Just by walking down the block, it is possible to be swept into its audio or video feeds, which can capture a nearly 360-degree view of their surroundings.

The new LinkNYC kiosks include a sizable 5G antenna, https://www.nytimes.com/2022/11/05/nyregion/nyc-5g-towers.ht...

> A curiously futuristic tower recently appeared on the corner of Putnam and Bedford Avenues in the Bedford-Stuyvesant neighborhood of Brooklyn. A gray column topped by a perforated casing, at a whopping 32 feet tall, it reaches higher than the three-story brick building behind it ... a cylindrical object roughly the size of a human being: a 5G antenna that is 63 inches tall and 21 inches in diameter, according to the company. It is accompanied by a box that is 38 inches high, 16 inches wide and 14 inches deep — about the size of a filing cabinet or a night stand.

Currently only 2009 LinkNYCs, according to https://data.cityofnewyork.us/Social-Services/LinkNYC-Map/tg...

My assumption is that these kiosks log all wifi and bluetooth traffic from nearby persons and vehicles, and USB device IDs of anything plugged in for charging. And of course the new ones will be logging 5G signals as well. Maybe there's not yet a centralized effort to coordinate these logs with the camera feeds to map people's physical appearance to their electronic footprints, but I can't imagine it isn't somewhere on the roadmap.

On one hand it might allow for softer and safer police practices. With ample video evidence and AI to recognize suspects, they can be apprehended at a time and place that’s safer for everyone.

End of the day this is the price being paid for unfettered rioting and the increase in petty crime since then. There’s just no way that is going to be allowed to happen again. Extensive camera networks and AI will deliver summary reports of top suspects that police will be assigned to apprehend with real-time location data sent to nearest units hours, days, or even months after the crime.

They might not get you for a few small shop lifting infractions but the cameras will watch and record and once you cross a threshold ($1000 In property?) you’ll make it into the summary report and face justice.

Cities in Mainland China have comparable density of cameras, if not more. And there's no mechanism to stop the police or the Chinese three letter agencies using them for facial recognition. Pretty scary.

Has anyone ever done research to see if pervasive use of facial recognition has any real effect on reducing crime?

It's not really about decreasing crime. It's really about increasing convictions (for better or worse).

This is the part I don’t understand. In the past decades surveillance tech has become more pervasive, and police funding has continued to increase. Yet the percent of homicides “cleared” by police has dropped from 90% to 50% over the past 50 years. All this surveillance, and somehow police continue getting worse at solving crimes

I’m curious as to what elements of policing have been the beneficiaries of increased funding. I’d love to see a budget and staffing breakdown of a police department from the 70s to now. I have a sneaking suspicion that it’s not the investigative side that’s grown hugely. I may well be wrong though, and just showing my bias.

It's possible that people are getting better at committing crimes at a faster rate than police are getting better at solving crimes. Maybe if we teleported today's average police department to the 1800s, the % of cleared homicides would be 100%. Who knows.

My guess is the increased evidence is lowering false convictions but it’s probably still very hard to find and prove the correct perpetrators.

> All this surveillance, and somehow police continue getting worse at solving crimes

Be careful with this line of thinking.

"cleared" generally means either an arrest was made (not a conviction) or they would've made an arrest but weren't able to (i.e. suspect died). A drop of a clearance rate just means it dropped. It only means cops in the past arrested people more often not that they actually arrested (or convicted) the correct people.

Increasing video coverage doesn't simply mean people are more likely to be imprisoned; it can also be used to prove innocence as well. Here's an example of a guy who was able to prove his innocence because he was captured on tape at a baseball game[1].

[1] https://www.theguardian.com/tv-and-radio/2017/sep/29/larry-d...

I’d expect increasing convictions and increasing the perceived likelihood of being convicted would both be negatively correlated with future crime levels.

Why? I mean, intuitively it makes some sense, but harsh penalties for drugs in the U.S. haven't really dented demand.

Anecdotally, NYC feels less safe than it has ever felt (in my 20+ yrs of being here). Most of the issues are not reportable crimes/would not be taken seriously by today's NYPD e.g. verbal harassment on the street, following, punching, etc. So while i'm sure i'm in the minority, i hope this leads to more policing.

London Sez:

"15,000 cameras? Heh. That's cute. Hold my lager."

(Hold my bitter more like, lager is more popular in the US :) )

Is the MTA doing this at its subway turnstiles? Everyday I see people just walk thru the open gates or jump right thru without paying. I feel like a chump for paying.

Reminded of Person of Interest

Pretty entertaining show actually.

I came here to say this. As time goes on I see more and more real tech that mirrors what is in the show, albeit without a controlling super-AI behind it all.

Palantir has access to quite a few databases, including health records, https://news.ycombinator.com/item?id=34149508

https://en.wikipedia.org/wiki/Domain_Awareness_System

Does seem like the west is essentially moving in the same direction as China on this topic, just less overtly

Another reason to mask up

we need cameras everywhere, this is a good start

Reply on news.ycombinator.com