Scott Berinato, CRN
Direct link to this page: https://www.hayadan.org.il/facercognition1.html
Many hopes were placed on facial recognition, especially after the events of 11/9. However, with the conclusion of two recent public trials of the technology, it seems that these hopes have been dashed on the ground of reality.
In the entertainment district of the city of Tampa in Florida, they began to test the recognition of facial features as part of the Super Bowl games. In 19 cases, facial features were identified against FBI databases but no arrest was made. In August, support for the program ended after some protesters wore Groucho Marx masks to prevent the system from recognizing their facial features.
In one case, according to St. Petersburg Times, police used a photo of a man eating lunch in the borough to demonstrate the system on the local news release. A woman in Oklahoma saw the photo and recognized the man pictured as her ex-husband who owed her child support for their son. After a police investigation, it was found that the man had never married.
If the Tampa trial was halted due to public opposition, it was poor performance that led to the termination of a second facial recognition trial conducted at Boston's Logan International Airport. Half of the terrorists who carried out the 11/9 attacks boarded the flights that crashed into the World Trade Center at this airport.
However, the tests carried out in Boston recently ended, after the system failed to recognize the facial features in 38 percent of the cases in which it was supposed to do so. While the false identifications by the human operators amounted to one percent of the cases, the rate of the automated false identifications reached fifty percent. The experiment was conducted between January 2002 and April 2002. The American Civil Liberties Union requested the results of the experiment under the Freedom of Information Act and published them in September 2003, reports the Boston Globe.
Referring to the final results of the experiment at Logan Airport, Meir Kakhtan, spokesperson for Identix - one of the companies that participated in the experiment, notes that many of the experiment's goals were achieved - among them the level of accuracy of the results. Kahten adds that this is an operational experiment conducted to find out if it is logistically possible to perform facial recognition at an airport and not a technical experiment designed to test the level of accuracy of the software.
However, it was the operational results that broke the law of the technology, according to the report compiled by the airport management. The report states that the program requires "much more involvement than expected" and due to the wrong identifications "a lot of hard work is required from the operators. They have to pay full attention to the machine at all times, so occasional breaks are necessary. At least two people must sit in each position." The high research and development costs slow down the rate of development of the technology and the "aggressive marketing strategies of the suppliers", the report states.
Charles Wilson, a scientist at the National Institute of Standards and Technology, who participated in tests conducted in 2002 by a supplier of facial recognition systems, mentions the experiment conducted at Logan Airport and another experiment that was stopped at Palm Beach International Airport in May 2002. However, according to him, this is not necessarily the case In the failure of the technology, but back to the ground of the reality of those who were carried away by the exaggerated promises of the marketers. The systems are not bad: they are just less good than many wanted to believe after 11/9.
"The optimism of the scientists is at a certain level, and that of the marketers is at a completely different level," says Wilson, adding that the conditions in which the system was operated must be taken into account: poor and uneven lighting, cameras that were placed high above the ground and therefore captured the facial features at a sharp angle - no wonder, According to him, the results were not impressive.
"Considering the results of the tests conducted at the airports, I believe that the systems operated as they were supposed to operate under these conditions," says Wilson. "To be honest, tests at airports were already conducted a year before the test at Logan, and the results were similar. The question arises why so much money is repeatedly wasted on experiments at airports? I think people were told they were getting a magic pill to fight terrorism. However, science cannot provide magic pills."
https://www.hayadan.org.il/BuildaGate4/general2/data_card.php?Cat=~~~687164334~~~143&SiteName=hayadan
