A suspect caught on camera could be identified in seconds due to new technology being tested by Leicestershire Police.
The NeoFace system compares measurements taken from an image of a face and compares it to the 92,000 on the force's database.
Officers said early results had seen a "high success rate of identification".
Concerns over privacy were rejected by senior officers who said a match did not constitute evidence.
Images could come from anywhere but CCTV or police body cameras had been the most common source so far.
Police said even digitised images of suspects had previously been compared manually, which could take many hours.
In early tests, about 200 suspect images had produced a 45% identification rate, which officers said was good.
Ch Insp Chris Cockerill said the system would not impact on civil liberties and could be a huge boost to the delivery of justice.
"We have these images anyway, all this is doing is making the search more efficient," he said.
"It could save tens of thousands of hours of work."
But Emma Carr, acting director of Big Brother Watch, said the technology should be used only with a "high level of accountability and oversight".
"Facial recognition cameras take the intrusiveness of CCTV to the next level, so it is absolutely essential that people are able to access meaningful redress when they feel their privacy is infringed," she said.
Leicestershire's database is made up of people who have either been arrested by police or have given their permission for their images to be stored, such as identity parade volunteers.
The trial is set to continue for six months after which results will be evaluated.
If successful it could be expanded across the East Midlands with each force's image database available for searches.