Citizen uses OpenCV to track speeders near his home • Hacker News Board

Re: Citizen uses OpenCV to track speeders near his home

#91

Post by bryanlarsen » Tue, Feb 02, 2016, 3:26 PM UTC

webjprgm wrote: post #80

I don't know if people share this opinion, but to me speeding is due to the inadequacy and inconsistency of speed limits. Near my home there is a major thoroughfare with a 35 MPH speed limit. It has no homes facing the road, just a few fenced backyards some fair distance away but mostly it has farms, ponds, and golf courses. People regularly go 40-50 MPH down this road. Another major road that crosses this one has a…

Sweden has a nice consistent policy on speed limits.

If pedestrians or bicyclists use a road, the speed limit is 20MPH, strictly enforced.

Consequently, they also have a grand total of 0 pedestrian fatalities most years.[1]

1: http://www.visionzeroinitiative.com/

Re: Citizen uses OpenCV to track speeders near his home

#92

Post by NetStrikeForce » Tue, Feb 02, 2016, 3:26 PM UTC

Someone1234 wrote: post #44

gus_massa wrote: post #31

I'll copy a very interesting comment by Anton Largiader http://www.cvilletomorrow.org/news/article/22908-locust-aven... > One very interesting approach I have seen overseas is a traffic light that turns red if you are speeding as you approach it. It's incredibly effective; instant feedback on your behavior, and immediate loss of any time you gained by speeding. No ticket, unless you run the light. Drivers quickly con…

I like that idea but I do worry about it causing rear-end collisions. In an ideal world drivers wouldn't speed and they also wouldn't tailgate, but in the world we currently occupy both are extremely common (even amongst law enforcement). So while this would act to reduce speeding, the unpredictability may increase accidents and it is definitely going to reduce traffic flow.

These are usually on blinking orange while you approach, warning that they might go red at any time. Most drivers unconsciously slow down and I haven't seen a rear-ending on these lights for more than 30 years.

Re: Citizen uses OpenCV to track speeders near his home

#93

Post by Tharre » Tue, Feb 02, 2016, 3:28 PM UTC

jurassic wrote: post #49

Some commenters here are commenting as if speeding is some petty thing that doesn't matter. Speeding kills. Something as small as 10 miles over a 30mph may mean a doubling in fatality rate for pedestrians and bicyclists. One source: https://www.aaafoundation.org/sites/default/files/2011Pedest...

Speeding kills ... your pocketbook: https://www.youtube.com/watch?v=2BKdbxX1pDw

Re: Citizen uses OpenCV to track speeders near his home

#94

Post by drzaiusapelord » Tue, Feb 02, 2016, 3:28 PM UTC

Someone1234 wrote: post #44

gus_massa wrote: post #31

I'll copy a very interesting comment by Anton Largiader http://www.cvilletomorrow.org/news/article/22908-locust-aven... > One very interesting approach I have seen overseas is a traffic light that turns red if you are speeding as you approach it. It's incredibly effective; instant feedback on your behavior, and immediate loss of any time you gained by speeding. No ticket, unless you run the light. Drivers quickly con…

I like that idea but I do worry about it causing rear-end collisions. In an ideal world drivers wouldn't speed and they also wouldn't tailgate, but in the world we currently occupy both are extremely common (even amongst law enforcement). So while this would act to reduce speeding, the unpredictability may increase accidents and it is definitely going to reduce traffic flow.

>I like that idea but I do worry about it causing rear-end collisions.

Chicago had a pretty significant red light camera deployment about six or seven years ago. Drivers used to be okay with blowing through a red that was a yellow a moment before in the past. Now everyone slams on their brakes when they see a yellow. According to time.com, "a 22% increase in injuries from rear-end accidents."

So yes, you can expect more rear end collisions and injuries when traffic light enforcement automation breaks up traditional driving habits.

Arguably, its okay to speed. Say doing 50 in a 40 when there's no traffic, like late at night. Why do we need a system to enforce this? I wish there was more fuzzy logic applied to traffic systems. Say a peak hours vs non-peak hours speed limit. Longer yellows during rush hour when everyone is stressed and potentially distracted. Instead we get hard and fast rules for all conditions, which doesn't make much sense.

http://time.com/3643077/red-light-cams-rear-end-collisions-c...

Re: Citizen uses OpenCV to track speeders near his home

#95

Post by cozzyd » Tue, Feb 02, 2016, 3:29 PM UTC

specialdragon wrote: post #39

In summary, he's made a speed camera and is now proposing the use of it to his local council. Are speed cameras not as ubiquitous in the USA as over here in the UK?

They're extremely controversial and unpopular, at least here in Chicago (where traffic enforcement is otherwise zero).

Re: Citizen uses OpenCV to track speeders near his home

#96

Post by pc86 » Tue, Feb 02, 2016, 3:32 PM UTC

acdha wrote: post #34

OhHeyItsE wrote: post #7

I wonder how the HN commentariat would feel about this if it was a government agency behind this rather than a fellow hacker...

The major question for me would be data retention – if they monitor speed (or failing to stop at a red light or stop sign, etc.) but fundamentally aren't doing anything different than if they parked an officer there with a radar gun, it doesn't bother me. Saving the data for later mining, OCRing license plates to build a record of cars traveling around town, etc. would be a different story because that affects everyo…

Unfortunately, for the following three scenarios:

  1 - Do nothing
  2 - Put a camera there just to get a % of cars going 5+ over the limit
  3 - Put a camera there to OCR everything and save it in perpetuity

The distance between 1 and 2 is quite large but the distance between 2 and 3 is very small and doesn't necessarily come with the same public notice/disclosure/outrage instituting 2 would. The safest course of action strictly from a privacy perspective is to oppose anything that makes implementing #3 easier.

Re: Citizen uses OpenCV to track speeders near his home

#97

Post by bryanlarsen » Tue, Feb 02, 2016, 3:33 PM UTC

drzaiusapelord wrote: post #94

Someone1234 wrote: post #44

Earlier quoted context omitted.

I like that idea but I do worry about it causing rear-end collisions. In an ideal world drivers wouldn't speed and they also wouldn't tailgate, but in the world we currently occupy both are extremely common (even amongst law enforcement). So while this would act to reduce speeding, the unpredictability may increase accidents and it is definitely going to reduce traffic flow.

>I like that idea but I do worry about it causing rear-end collisions. Chicago had a pretty significant red light camera deployment about six or seven years ago. Drivers used to be okay with blowing through a red that was a yellow a moment before in the past. Now everyone slams on their brakes when they see a yellow. According to time.com, "a 22% increase in injuries from rear-end accidents." So yes, you can expect m…

You omitted the fact that it reduced injuries by 15% from side-swipe collisions.

The Time omitted the fact that side-swipe collisions tend to have more serious injuries than rear-end collisions. They also failed to mention that red light cameras reduce the number of pedestrian and bicycle collisions.

Re: Citizen uses OpenCV to track speeders near his home

#98

Post by drzaiusapelord » Tue, Feb 02, 2016, 3:35 PM UTC

bryanlarsen wrote: post #97

drzaiusapelord wrote: post #94

Earlier quoted context omitted.

>I like that idea but I do worry about it causing rear-end collisions. Chicago had a pretty significant red light camera deployment about six or seven years ago. Drivers used to be okay with blowing through a red that was a yellow a moment before in the past. Now everyone slams on their brakes when they see a yellow. According to time.com, "a 22% increase in injuries from rear-end accidents." So yes, you can expect m…

You omitted the fact that it reduced injuries by 15% from side-swipe collisions. The Time omitted the fact that side-swipe collisions tend to have more serious injuries than rear-end collisions. They also failed to mention that red light cameras reduce the number of pedestrian and bicycle collisions.

The article certainly mentions the holistic view here:

"Taken together, the study shows a statistically insignificant increase of injuries by 5%."

Time's source is the Tribune:

http://www.chicagotribune.com/news/watchdog/redlight/ct-red-...

Emanuel has credited the cameras for a 47 percent reduction in dangerous right-angle, or "T-bone," crashes. But the Tribune study, which accounted for declining accident rates in recent years as well as other confounding factors, found cameras reduced right-angle crashes that caused injuries by just 15 percent.

At the same time, the study calculated a corresponding 22 percent increase in rear-end crashes that caused injuries, illustrating a trade-off between the cameras' costs and benefits.

The researchers also determined there is no safety benefit from cameras installed at intersections where there have been few crashes with injuries. Such accidents actually increased at those intersections after cameras went in, the study found, though the small number of crashes makes it difficult to determine whether the cameras were to blame.

>They also failed to mention that red light cameras reduce the number of pedestrian and bicycle collisions.

From what I can tell, no one has done a study on that and it sounds questionable to me to see any results there. In fact, Chicago has a big problem with cyclists hitting pedestrians since it added so many bike lanes. But that's a discussion for another time I suppose and a problem that'll probably take care of itself as cyclists and pedestrians get used to each other.

Re: Citizen uses OpenCV to track speeders near his home

#99

Post by pc86 » Tue, Feb 02, 2016, 3:36 PM UTC

lorenzhs wrote: post #78

otoburb wrote: post #35

Earlier quoted context omitted.

Generally it seems you may photograph and film public property, assuming you believe Wikipedia.[1] [1] https://en.wikipedia.org/wiki/Photography_and_the_law#Public...

This wouldn't fly at all in Germany. If your surveillance camera films public ground, even if it's not the primary object being photographed, you're acting illegally. If, for example, you're trying to monitor your front door to catch a potential burglar and you can see the pavement and the people on it on the video, that's not allowed. You won't be able to use the footage in court even if it shows the burglar clearly…

This seems bizarre to me. Why would you have any expectation of privacy walking down a public street?

Re: Citizen uses OpenCV to track speeders near his home

#100

Post by notacoward » Tue, Feb 02, 2016, 3:36 PM UTC

pc86 wrote: post #69

notacoward wrote: post #46

Earlier quoted context omitted.

I wasn't suggesting anything that triggers red lights or leads directly to a citation. Just gathering the data would often be useful, to drive design of intersections and allocation of other resources in a quantitative way. Also, if people aren't maintaining a safe distance from the car in front of them, then that's useful information too. So is video of people illegally using their celphones while driving, which thi…

Well if you rear end someone you will be cited for it 99% of the time. But surely your answer to something that quantifiably increases injury rates isn't "who cares it's their own fault" is it? But since you asked for a citation: http://bfy.tw/4343

Putting words in someone else's mouth is a tactic commonly used by those who can only formulate stupid statements. Of course I care about things that increase or decrease injury rates. That's why I would like to encourage safer behavior. You should haved added "debunked" to your snarky citation, and you might have found this.

http://www.cnet.com/roadshow/news/red-light-camera-myths-deb...

Also, as I said, if people are tailgating then that's a separate problem that can also be addressed by this technology. It's insane to say that we shouldn't discourage one dangerous behavior because of some other equally-addressable dangerous behavior. People respond to the knowledge that they're more likely to be caught when they do something wrong or stupid. An actual police officer issuing citations for exactly these offenses is effective, but costly. Having the information about frequency of offense at different locations allows those scarce resources to be allocated to greatest effect. What's your problem with that?