alimaui
Well-known member
- Joined
- Mar 23, 2002
- Posts
- 207
Machines at an airport fail face-recognition test
By LANCE GAY
Scripps Howard News Service
May 16, 2002
- Random tests of facial-recognition technology at a Florida airport this spring show that it failed to identify even half of the employees who went through screening.
The American Civil Liberties Union released the results of the first four weeks of an eight-week study at Palm Beach International Airport. According to the group, the system made a match only 455 out of 958 times that volunteers who had their faces recorded digitally in the computer system went through it.
"The face-recognition technology was less accurate than a coin toss,'' said Barry Steinhardt, director of the ACLU's technology program, which is campaigning against the machines. Steinhardt noted the Palm Beach results come in the wake of failures of such technology to identify people when the machines were used at the Super Bowl, and on the streets of Tampa, Fla.
The ACLU obtained the results through Florida's sunshine laws.
Visionics, the Jersey City, N.J., company that makes the machines, contended that poor lighting at the airport was responsible for the poor results, and that the ACLU did not obtain data from three other airports - Boston's Logan International, Dallas-Fort Worth and Yosemite International in Fresno, Calif. - that had far better success rates.
"The project is in its test phase and we are adjusting things," said Visionics spokesman Meir Kahtan. He said the results from Logan and Dallas-Fort Worth showed a success rate of better than 90 percent, and that Fresno's success rate was 80 percent.
"The ACLU puts out information that suits its own agenda," he said. Kahtan said he did not know when the final results from all tests would be released.
Facial-recognition, or biometric, technology relies on reducing the measurements on human faces to mathematical formulas. The information is then matched with formulas of pictures stored in a computer database.
Privacy advocates oppose the machines, saying they only serve as a deterrent but aren't sufficiently reliable to identify dangerous people like terrorists, even if there are good up-to-date pictures of their faces.
John Magaw, director of the new Transportation Security Administration that oversees airport security, says he is weighing using the machines at all U.S. airports as part of an effort to improve security and identify suspected terrorists before they can board U.S. aircraft.
Bruce Pelly, director of Palm Beach International Airport, said he did not regard the tests as a success.
"It is not satisfactory. We had hoped to get a higher percentage,'' he said. The ACLU obtained data only from the first phase of the tests, but Pelly said the second testing period produced similar results.
The Boston machines got better results because employees there looked into the cameras, Pelly said. "Ours was more random."
Test results found that the machines had problems identifying people wearing glasses. "Tinted lenses diminished the system's effectiveness," the report on the first phase says. There was also "a substantial loss of matching" if the face of a test subject was caught at a side angle, or was looking down or up. The machines also registered an average two or three "false positives," or false alarms, an hour where people were identified improperly.
"There are a lot of issues using this equipment," Pelly said.
Randall Marshall, legal director of the ACLU in Florida, said the Palm Beach results show that facial technology "is a clunker that holds little promise to make us safer."
On the Net: www.aclu.org
FROM DRUGEREPORT.COM
By LANCE GAY
Scripps Howard News Service
May 16, 2002
- Random tests of facial-recognition technology at a Florida airport this spring show that it failed to identify even half of the employees who went through screening.
The American Civil Liberties Union released the results of the first four weeks of an eight-week study at Palm Beach International Airport. According to the group, the system made a match only 455 out of 958 times that volunteers who had their faces recorded digitally in the computer system went through it.
"The face-recognition technology was less accurate than a coin toss,'' said Barry Steinhardt, director of the ACLU's technology program, which is campaigning against the machines. Steinhardt noted the Palm Beach results come in the wake of failures of such technology to identify people when the machines were used at the Super Bowl, and on the streets of Tampa, Fla.
The ACLU obtained the results through Florida's sunshine laws.
Visionics, the Jersey City, N.J., company that makes the machines, contended that poor lighting at the airport was responsible for the poor results, and that the ACLU did not obtain data from three other airports - Boston's Logan International, Dallas-Fort Worth and Yosemite International in Fresno, Calif. - that had far better success rates.
"The project is in its test phase and we are adjusting things," said Visionics spokesman Meir Kahtan. He said the results from Logan and Dallas-Fort Worth showed a success rate of better than 90 percent, and that Fresno's success rate was 80 percent.
"The ACLU puts out information that suits its own agenda," he said. Kahtan said he did not know when the final results from all tests would be released.
Facial-recognition, or biometric, technology relies on reducing the measurements on human faces to mathematical formulas. The information is then matched with formulas of pictures stored in a computer database.
Privacy advocates oppose the machines, saying they only serve as a deterrent but aren't sufficiently reliable to identify dangerous people like terrorists, even if there are good up-to-date pictures of their faces.
John Magaw, director of the new Transportation Security Administration that oversees airport security, says he is weighing using the machines at all U.S. airports as part of an effort to improve security and identify suspected terrorists before they can board U.S. aircraft.
Bruce Pelly, director of Palm Beach International Airport, said he did not regard the tests as a success.
"It is not satisfactory. We had hoped to get a higher percentage,'' he said. The ACLU obtained data only from the first phase of the tests, but Pelly said the second testing period produced similar results.
The Boston machines got better results because employees there looked into the cameras, Pelly said. "Ours was more random."
Test results found that the machines had problems identifying people wearing glasses. "Tinted lenses diminished the system's effectiveness," the report on the first phase says. There was also "a substantial loss of matching" if the face of a test subject was caught at a side angle, or was looking down or up. The machines also registered an average two or three "false positives," or false alarms, an hour where people were identified improperly.
"There are a lot of issues using this equipment," Pelly said.
Randall Marshall, legal director of the ACLU in Florida, said the Palm Beach results show that facial technology "is a clunker that holds little promise to make us safer."
On the Net: www.aclu.org
FROM DRUGEREPORT.COM