A future with sophisticated robotics and artificial intelligence might be closer than we think, and some of them are already solving crime. It appears a machine based algorithm has just located the Capital Gazette suspect among a host of a whopping 10-million photos – a world first for facial recognition software and familiar devices.
This remarkable feat was courtesy of law enforcement personnel from Maryland, who used facial recognition software to identify Jarrod Ramos, the suspect of the recent shooting at Capital Gazette that unfortunately left one newspaper sales associate and four (4) journalists dead.
When Ramos had been apprehended in Annapolis at the scene of the crime, he didn’t seem to have any intention of speaking to the police, and didn’t have any identification with him. As a result, police had actually taken a photo of the then-unidentified Ramos and uploaded it into the state’s sophisticated Maryland Image Repository System (MIRS).
The MIRS, which is a database with a whopping 10-million mugshots of people and driver’s license images, was then been able to get a “hit” for 33-year-old Jarrod Ramos. According to the MIRS, it appears Ramos is from Laurel, Maryland who may have had a grudge against the local paper that lasted for years.
The Baltimore Sun reported that this feat of identification isn’t exactly new, given that the MIRS can be accessed by as much as 6,000 to 7,000 officials for 175 times in just one week.
Maryland’s Public Safety and Correctional Services secretary Stephen Moyer said this feat of facial recognition is “performed as designed.” He added this proves MIRS’ continuation of being quite a valuable asset when it comes to fighting crime in Maryland.
Interestingly, it was revealed that the MIRS wasn’t exactly Maryland’s first choice when it came to identifying Ramos. Timothy Altomare, County Police Chief (Anne Arundel) said their initial attempt to identify him via fingerprints had failed because of “lag” in the system, suggesting there might be errors or malfunctions in the technology.
The above is plausible, given fingerprinting errors appear common. Some industry estimates also put the error rate to be as high as 10-percent, based on factors such as poor fingerprinting technique or equipment, but also the prints. Anyone’s fingerprints are only as sturdy as their skin, which means extreme swelling or manual labor may make them unreadable.
Some experts try to avoid this mistake by making sure multiple fingers are printed all at once, or using other techniques that can deal past abrasion. Unfortunately, the ongoing debate towards reading prints from damaged fingers has yet to find an optimal solution.
Meanwhile, aside from the 10-million photos the MIRS has access to, it’s also known to have access to the FBI’s database of mugshots, which adds yet another 25-million to sift through. Given Ramos was actually previously convicted for harassment, it’s likely that his mugshot may have been identified with the FBI’s servers.
Of course, regardless of the success in identifying Ramos, a lot of concerns on civil liberties are still there with regards to the system. Altomare also admitted the system had been criticized by advocates of data privacy, as police have the responsibility of “removing” people who were arrested but were eventually found innocent. Unfortunately, given that facial recognition software isn’t exactly audited all the time, there’s no guarantee that his is actually happening.
You May Also Like: