Smart Cameras Blog

[From Red-light camera should capture right car with right plate :: Naperville Sun via Instapundit and Gapers Block ]

Consider the plight of this senior citizen who received a red light camera ticket for a car that wasn’t hers.

After many phone calls, insisting this was not her vehicle, nor was she anywhere near the location indicated, she was instructed to protest the ticket at the Daley Center.A protest filing fee was paid. Three appearances were required in traffic court. Each one became more frustrating than the one before. It seemed the protester was not innocent before being proved guilty but quite the contrary. It became obvious that no representative of the court believed the story.The third appearance finally got the court’s attention because the license plate was produced as evidence, and although it did match the numbers on the plate, it was not identical to the plate on the photo.After all the dust had cleared and the expenses were paid by this tenacious senior citizen, including time, parking fees, tolls as well as anguish and distress, the case was dismissed.It was dismissed simply because, after investigation by this lady, it was discovered that two vehicles registered in the state of Illinois can have the same identical plate number. The only difference is that one is a vanity plate.

This story is a marked contrast to the piece on RedFlex in the Tribune a few weeks ago. It described the process for ticketing cars:

Once through the lobby, visitors come upon a half-dozen or so violation reviewers, who watch every single 12-second video clip and still image that comes to RedSpeed showing a possible violation. Because drivers regularly trigger the cameras without a violation occurring—often by stopping just past the stop line or stopping abruptly after speeding—it’s up to the reviewers to determine whether they think there’s a violation.If they think not, the file gets purged; but if they believe a citation should be issued, the reviewers send the information to a senior reviewer. If that person agrees, the file gets forwarded to the respective law-enforcement agency, which decides whether to issue a citation.

This story points out a disconnect between the process of identifying a license plate and confirming that it matches the vehicle description associated with the license plate. Part of this is a technical problem. It was reported last year that the tollway Ipass cameras couldn’t detect vanity plates properly. Apparently, this is also a problem for red light cameras. The second is a bureaucratic issue. It is well known there is a problem with vanity plates. 25% of the plates in Illinois are non-standard. Why aren’t there measures in place to prevent this from occurring? Here is a graphic from the Daily Herald on vanity plates.

rshah Chicago, Red Light Cameras

[From Taipei Times - archives and Taiwan News and China-Taiwan News]

Taipei Mayor Hau Lung-bin announced on April 8 that his government would spend NT$1.6 billion (US$50 million) to install 13,000 “intelligent,” high-performance cameras in the city. Of the 13,000 cameras, 1,500 will be high-resolution one-mega-pixel models. The city is replacing 12,000 existing security cameras.

Here are the scenarios for the smart cameras:

Closed-circuit television (CCTV) technology is so far advanced nowadays that it’s possible to search for certain actions in recorded video. For example, when investigating an ATM crime, police can set a search condition for “a person staying near a machine for longer than 10 minutes” and the system will find all video meeting the criterion.

It’s also possible to have the system to alert police automatically when a car matching a previously entered description appears in the camera lens, helping to solve car thefts more quickly.

The article notes there are still concerns about privacy and how footage will be used.

Asked whether the new equipment would mean a further erosion of residents’ privacy, Hau said that no one will be given access to the recorded tapes without a justifiable reason. He added that all the video footage will be recordings of activities on public roads and will not infringe on people’s privacy in their homes.

According to police executives, the new cameras, linked to a computer terminal, will be able to home in on suspected targets and will have the ability to detect cars thefts by reading the number plates of vehicles and sounding alarms.

rshah Other Cities

Lucas D. Introna and Helen Nissenbaum have put together a comprehensive look at facial recognition. The title of the report is Facial Recognition Technology: A Survey of Policy and Implementation Issues.

The report:

highlights the potential and limitations of the technology, noting those tasks for which it seems ready for deployment, those areas where performance obstacles may be overcome by future technological developments or sound operating procedures, and still other issues which appear intractable. Its concern with efficacy extends to ethical considerations.

For the purposes of this summary, the main findings and recommendations of the report are broken down into five broad categories: performance, evaluation, operation, policy concerns, and moral and political considerations.

Introna also wrote a more scholarly paper a few years ago on facial recognition titled, Picturing Algorithmic Surveillance: The Politics of Facial Recognition Systems.

rshah Applications, Facial Recognition, Policy

When a camera is used, there maybe certain parts within a camera’s field of view that need to be kept private. The most obvious is if a camera could see into a private area, such as a window or doorway to a house. If this occurs, the appropriate measure is to use masking, privacy zones, or blanking. This can be done physically, by limiting the camera’s field of view during the installation process. This can also be done by software, however, software creates a risk that “someone might, somehow, get access to the video prior to masking, or perhaps turn off the masking” according to Sightmind.

One example of software is Pelco’s window blanking system. It is capable of blocking of specific areas, such as open windows to satisfy privacy issues. “In Spectra III SE, you will be able to define up to eight, four-sided polygons, with sides of any length, and turn on the blanking at specific zoom ratios.”

I don’t know how effective these systems are in practice, but I think most police departments should be using some type of masking system when setting up cameras near residential areas. As a further measure, policy departments should also be keeping system logs concerning what cameras are looked at by what user and when.

Here are some images of masking at work:

rshah Policy

[From In Hard Focus: Manchester Airport Downgrades Matching Threshold on Facial Scanners]

Here is the money quote from the Telegraph article that was posted:

Airport face scanners designed to verify travellers’ identity against their passport photographs are working at such a low level that they would be unable to tell the difference between Osama bin Laden and the actress Winona Ryder, it has been claimed.

Here is more detail:

In a leaked memo, an official says the machines have been recalibrated to an “unacceptable” level meaning travellers whose faces are shown to have only a 30 per cent likeness to their passport photographs can pass through.

The machines, undergoing trials at Manchester airport, have apparently been questioning so many passengers’ identities that they were creating huge queues. The technology was designed to help immigration officials spot people traveling under false passports, particularly terrorists, but the multi-million pound scheme now appears to be in jeopardy.

Rob Jenkins, an expert in facial recognition at Glasgow University’s psychology department, said lowering the match level to 30 per cent would make the system almost worthless. Using facial recognition software from Sydney airport in Australia set at 30 per cent, he found the machines could not tell the difference between Osama bin Laden and the actors Kevin Spacey or even the actress Winona Ryder while Gordon Brown was indistinguishable from Mel Gibson.

Stephen Russell adds a comment:

I’ll be interested to see if an official response to this “leaked memo” is released that explains the drastic drop in standards and their reason for the large number of false-positives. It could be the result of their technology choice, or just the result of poor, old, and varied photos common to passports. Likely both.

rshah Applications, Facial Recognition

[From 10 facts about red-light cameras -- chicagotribune.com]

Here is a summary of how the red light cameras operate in Chicago. This material was compiled from a number of sources. The vendors for Chicago’s cameras are RedSpeed Illinois and Redflex Traffic Systems.

The system operates by using “magnetic sensors installed at various points leading up to the stop line to determine whether the camera should begin taking pictures and video. (Video clips are about 12 seconds.) There are no sensors past the stop line, meaning if you’re in the middle of the intersection—waiting to turn left—and the light turns red, the camera should not record the incident. The average intersection has two cameras, although the number ranges from one to four.” (from Tribune).

This technology is also known as loop detectors. Once triggered the cameras take 3 images and 12 seconds of video of each violation. A violation occurs when a vehicle enters an intersection 1/10 of a second after the light has changed to red. The only two legitimate reasons for disobeying a red light are participating in a funeral precession or yielding to an emergency vehicle.

Loop detection technology was chosen, even though its more expensive, because the city felt video detection is not yet mature and has a high failure rate. The cameras are are fixed so they cannot be remotely maneuvered.

A bit more detail can be found at the web site for RedSpeed Illinois. Their explanation is:

The state of traffic signals is continuously monitored and their ‘on’ time recorded by the system. When the red aspect is detected, a timer is started. When this timer exceeds the programmed ‘time into red’, the two road loops are monitored for detection of a moving vehicle. If a vehicle is detected, the program proceeds to violation capture, then two wide-angle images and one zoom image of the vehicle are recorded.

However, the next generation system is known as RedSpeed. Here is what it can do:

The system detects speed violations during the green and amber phases, and also detects red light violations. Three images and a 12 second video clip are captured. The first is a zoom shot showing the vehicle number plate. The second is a wide angle image including the make, model, and colour of car and the offence data. The system then delays, for a period related to the vehicle’s speed, and captures a third frame, also a wide angle image showing the time delay between frames. The two wide-angle images can be used as evidence to show that the vehicle was moving at the time of the recording. (My emphasis)

I would not be surprised to see the RedSpeed cameras deployed next. We will hear how speeding through intersections causes many accidents and how it is necessary to use cameras to deter this behavior. This new technology can save lives and raise money for cities. (They can also be used to gather lots of data on us, see here and here.)

rshah Chicago, Red Light Cameras

[From Are Indy's police surveillance cameras worth it? | IndyStar.com | The Indianapolis Star ]

Indianapolis has a small camera network, I first mentioned it last year. For background, they have spent over a million dollars and are at 54 cameras with another 40 coming online this year.

The problem noted in the article is the city doesn’t keep any data on the effectiveness of the cameras. As a result, taxpayers and police don’t know if these cameras are really helping to address crime. The article notes recent studies show cameras have limited effectiveness, e.g. San Francisco. The article also states that Chicago “reported that neighborhoods with cameras operating for more than six months saw a 30 percent decrease in crime and a 60 percent drop in drug incidents.” However, Chicago, unlike San Francisco, has not made public its data or analysis for these statistics.

The lack of data will probably not hamper the growth of camera networks. Unlike many other technologies, people believe in their guts that cameras can make a difference. In the absence of data, they will probably prefer cameras. However, for the policy wonks and those that care about the long term, the lack of data will mask the true value of cameras. We won’t know if money should be spent on cameras, officers, other police technologies, or for other strategies.

rshah Chicago, Other Cities, Policy