This article is more than 1 year old

What could go wrong? Delta to use facial recog to automate bag drop-off

Just make sure you still resemble your passport photo

Delta Air Lines plans to deploy four self-service bag drop machines at Minneapolis–St Paul International Airport this summer, one of which will include a facial recognition system to match those depositing bags with their passport photos.

Delta senior veep of airport customer service and cargo Garth Joyce, in a canned remark, characterized the airline's $600,000 investment in automation as a way to save customers time. "Since customers can operate the biometric-based bag drop machine independently, we see a future where Delta agents will be freed up to seek out travelers and deliver more proactive and thoughtful customer service," he said.

Joyce did not clarify why automating customer interaction would necessarily lead airline employees to take more initiative or to become more considerate, particularly when it might just as well allow Delta to employ fewer people.

Delta claims studies indicate that self-service bag drops have the potential to process twice as many customers per hour as those staffed by employees.

A Delta spokesperson in a phone interview with The Register explained that the facial recognition system would be available to national and international travelers using passports as identification. The motivation, she said, was to improve the customer experience.

The self-service bag drop provides customers with RFID-embedded baggage tags, introduced last year, that provide travelers with visibility into the location of their bags across the 84 largest US airports served by Delta.

According to Delta's spokesperson, "For last year, there was a 3 per cent improvement in [mishandled baggage reports] and the RFID mechanisms were installed throughout the year – we expect a 10 per cent improvement overall."

Various academic studies and recent government reports [PDF] document the potential inaccuracy of facial recognition systems. A NIST study [PDF] published in March found that facial recognition systems in a boarding gate scenario misidentify 6 per cent of the people in a 480-person data set, and 18 per cent of people in a 48,000-person data set.

Nonetheless, the technology can be expected to become more common, in part due to a 2004 legislative requirement to expand the use of biometric identifiers. That mandate has led US Customs and Border Protection (CBP) to explore various biometric initiatives.

"Over the course of the next several months, CBP will expand the deployment to seven additional airports to continue biometric exit implementation," a CBP spokesperson told The Register in an email.

"Under this approach, CBP will learn best practices for operations and integration into existing airline boarding processes. CBP is working closely with stakeholders to ensure successful implementation of biometric exit, transform the entry process, and expand public-private partnerships."

Last year, from June through September, CBP worked with Delta to test facial recognition technology at Hartsfield–Jackson Atlanta International Airport. The test covered a single daily flight from the US to Japan.

CBP presently uses facial recognition technology at John F Kennedy International Airport and Washington Dulles International Airport. It also conducts biometric verification on a flight to Mexico City, Mexico.

"CBP sees potential for the technology to transform the travel process, provided privacy issues can be addressed," CBP's spokesperson said, citing the opportunity to reduce the need to produce documents while traveling.

Addressing privacy issues became a lot easier for the CBP earlier this year, thanks to a White House Executive Order that directed all US government agencies to ensure that their privacy policies "exclude persons who are not United States citizens or lawful permanent residents from the protections of the Privacy Act regarding personally identifiable information." ®

More about

TIP US OFF

Send us news


Other stories you might like