Police bandit duel, face recognition really so useful?

Phoenix Technology News, Beijing time on January 13, face recognition technology has become the iPhone and other high-end mobile phone standard, it has also become a tool to assist the police to solve the case. The New York Times published today revealing how the Florida police used the face recognition system to identify suspects, and what are its short boards?

Police bandit duel, face recognition really so useful?

U.S. police use face recognition technology to solve cases

Here’s the full text of the article:

North of Orlando, Florida, police broke the tires of a stolen Dodge Magnum and pulled it down after a high-speed chase. They arrested the driver, but could not confirm his identity. The man, who had no identity card, fainted after stuffing something into his mouth. Police said he appeared to have bitten off his fingerprints.

Investigators were forced to turn to one of the county’s oldest and largest face recognition systems: a statewide face recognition program that began operating in Pinellas County nearly 20 years ago. Twenty years ago, law enforcement was just beginning to use face recognition technology.

Police searched a large database of photos of the man and found a possible match. The 2017 case is the first of more than 400 successful suspect identification cases by the Pinilas County Facial Recognition System since 2014.

The New York Times review of these Florida facial recognition records is the most comprehensive analysis to date of a local law enforcement agency’s face recognition system, giving a glimpse into the potential and limitations of facial recognition technology.

The potential and limitations of face recognition

Florida State Police say they check the system 4,600 times a month. However, face recognition technology is not a panacea. The documents show that only a small number of inquiries can crack the police’s public investigation of unidentified suspects. When the image is clear, the face recognition tool effectively identifies the detainee who does not cooperate because the suspect uses a fake ID and photo from an anonymous social media account. But as investigators try to find suspects who appear briefly in grainy surveillance video footage, the face recognition system takes a big toll.

At the same time, the Florida Face Recognition Project highlights concerns that new technology could violate due process. Documents show that the system operates with little oversight and its role in legal cases is not necessarily disclosed to the accused. Although police say investigators don’t rely on facial recognition results to issue arrest warrants, documents show that police sometimes struggle to gather other evidence.

“It’s peddled that it’s accurate enough to do all sorts of crazy things,” said Clare Garvie, a senior fellow at georgetown University School of Law’s Center for Privacy and Technology. “

In recent years, while face recognition has become a daily tool for phone unlocking and social media tagging photos, it has also caused controversy. The field has attracted latecomers such as Amazon and is used by law enforcement in New York, Los Angeles, Chicago and elsewhere, as well as by the FBI and other federal agencies. Amazon is also promoting face recognition technology to the police. Research on facial recognition systems is still scarce, but a 2016 study found that half of U.S. adults have been included in a law enforcement facial recognition database.

Police argue that face recognition makes the public safer. But a handful of cities, such as San Francisco, have banned law enforcement from using facial recognition tools because of concerns about user privacy breaches and the erroneous matching of face recognition technology. Civil liberties advocates have warned that facial recognition technology could be used maliciously.

In Florida, face recognition has long been part of routine policing. Twenty years ago, the Pinilas County Sheriff’s Office tried federal funds to try face recognition technology. Now, it has actually become Florida’s face recognition service, accessing more than 30 million images, including driver’s licenses, suspect photos, and teen appointment photos.

“People think it’s new,” Said Bob Gualtieri, the Pinillas County Sheriff, referring to facial recognition technology. “

Can face recognition alone do it?

So far, only one U.S. court has ruled on law enforcement’s use of facial recognition, increasing the belief that defendants’ right to know is limited.

In 2015, Willie Allen Lynch was charged with selling $50 million worth of cocaine after the Pinellas County face recognition system suggested he was a possible suspect. Lynch argued that he was mistaken for a suspect and sought other images that might match the suspect. The Florida Court of Appeals rejected his request and sentenced him to eight years in prison.

Any technical findings that are presented as evidence should be analysed through special hearings, but the facial recognition results have never been considered reliable enough to withstand such challenges. However, face recognition results can still play a significant role in investigations until more reliable forensic techniques are reviewed by the judiciary.

The laws and court rules of each u.S. state vary on which investigative material sits to be shared with the defendant. This has led some law enforcement officials to argue that they do not need to disclose the use of facial recognition.

In some Florida cases, facial recognition technology was not mentioned in the initial arrest warrant or court filings. Instead, detectives identified “investigative techniques” or “confirmation attempts” in court documents and listed the cases as a success in the Face Recognition app in Pinellas County records. Defense attorneys said in an interview that the use of face recognition is sometimes mentioned at the end of the discovery process, but not necessarily.

Aimee Wyant is a senior assistant public defender for the Judicial Circuit In Pinellas County. She said defense lawyers should be aware of all the information used in the investigation. “Once the police find the suspect, they’re like a dog with a bone, and that’s their suspect,” Wynt said. “

Law enforcement officials in Florida and elsewhere stressed that face recognition technology should not be relied upon to carry out arrests. “Computer-matching technology alone cannot arrest anyone,” James O’Neill, the new police commissioner, said in June.

In most Florida cases, investigators followed similar guidelines. But in a few cases, face recognition was the main basis for police arrests, according to court documents. Last April, for example, a Tallahassee police officer investigating the theft of an $80 cell phone obtained a store surveillance image and found a possible match in the face recognition system. She wrote in court documents that she looked back at the surveillance video and identified the suspect.

A police spokesman suggested that the suspect’s identification had been verified by the face recognition system. “We’re not just saying ‘it’s him’ or even going to investigate,” she says. “This case is under consideration.

“Roll Call Game” Into History

The launch of the Face Analysis Comparison and Inspection System (FACES) in Pinellas County began with a $3.5 million federal grant arranged by Rep. Bill Young in 2000. Bill Young, a Florida Republican, heads the House Appropriations Committee.

Law enforcement in other states have not had an early test of facial recognition systems, such as California, which has relied on face recognition systems to arrest only one person in four years. However, the potential of face recognition technology remains attractive. Pinilas County’s initial plan for face recognition technology is to use it in the current prison’s facial photo system. After 9/11, the FACES program was expanded to the airport. Eventually, Pinilas County police were able to upload photos taken with digital cameras to the system while on patrol.

The Pinellas County Sheriff’s Office says the face recognition program received more than $15 million in federal grants before the county began covering annual maintenance costs in 2014 and is now about $100,000 a year.

The first arrest in Florida under the Face Recognition System occurred in 2004. According to local media reports at the time, a woman being hunted for violating probation gave the police a false name. As the use of FACES expanded across the state, the image library incorporated the driver’s license system and arrests increased. By 2009, the Pinilas County Sheriff’s Office had attributed nearly 500 arrests to the face recognition system. By 2013, that number was approaching 1,000. Only a few cases have been made public.

The latest data show that the Pinilas County Facial Recognition System has successfully identified suspects in more than 400 cases since 2014. However, this data is flawed: not all successful identification cases are documented, and questions or negative results are not recorded. However, together with relevant court documents, about half of the case records are easily accessible. The latest list of face recognition cases also analyzes which crimes that face recognition best assists in solving: shoplifting, counterfeiting of cheques, and identity fraud.

In this series of cases, the police are seeking identity checks. “We call it a ‘name game,'” Sheriff Gu said. “

In some 36 court cases, face recognition played an important role, despite blurred images. Of these, nearly 20 cases involved petty theft, while others were more serious.

Pinellas County documents show investigators used facial recognition technology to lock down the suspect after a gun robbery at an ATM near Hillsborough County in 2017. They showed surveillance video of the ATM to the suspect’s girlfriend, who identified the suspect. The suspect later pleaded guilty.

In nearly 20 Pinellas County facial recognition cases, investigators are trying to identify people who do not know who they are, such as Alzheimer’s patients and murder victims. The Pinilas County Sheriff’s Office says face recognition technology is sometimes used to assist in identifying witnesses.

Some of the most cutting-edge applications of face recognition technology in the field of crime detection have not shown positive results and have been abandoned, such as in the use of airports.

Image quality is key

“It depends on the image quality,” said Jake Roberto, a technical support specialist with the Pinilas County Sheriff’s Office who helps run the face recognition system. “

The FACE system was developed by the French company Idemia, whose prototype algorithms have also excelled in recent tests at the National Institute of Standards and Technology. However, the FACE system used by law enforcement does not necessarily contain the latest algorithms. For example, the FACE system used in Pinilas County was heavily updated in 2014, but the county has been evaluating other updates. Idemia declined to comment.

In recent years, the upgrade of excellent face recognition technology has been alarming. In government tests, face recognition algorithms were able to compare with a database of 1.6 million facial photos. In 2010, the error rate of face recognition systems under ideal conditions was just under 8%. Ideal conditions here are adequate lighting, high resolution, front-face photos. By 2018, the error rate had fallen to 0.3%. But under surveillance video conditions, law enforcement can’t expect face recognition to be so reliable.

Perhaps the biggest controversy with face recognition technology is its erratic behavior in identifying people of different races. Test data released by the National Institute of Standards and Technology in December showed that the type of face recognition system used in government surveys was more likely to be error-prone when evaluating images of black women. Florida law enforcement officials say the technology’s performance does not indicate a degree of racial bias.

Officials in Pinilas County and elsewhere also emphasized the role of manual audits. However, tests using passport images have shown that it is difficult for human auditors to identify the right person in similar face recognition results. In these tests, passport system employees often choose the wrong one.

Poor quality images result in matching errors. Dim lighting conditions, a certain angle of the face, and light camouflage with a baseball cap or sunglasses can affect the accuracy of face recognition.

In some countries where civil liberties are more legal, the short board of facial recognition technology is exposed, especially when used to find criminals in the crowd. London has a huge network of CCTV cameras, but a facial recognition study found that the system was tested 42 times, but only eight were proven accurate.

Former and current officials in Pinilas County said they were not surprised by the results. “If you want to investigate bank robberies and convenience store robberies, face recognition technology doesn’t work,” said Jim Main, who was in charge of the Facebook N.H.L. job and retired in 2014. “