Thunder Bay

Eye on the Street software upgrade uses AI to speed up searches in Thunder Bay

Thunder Bay police say a new software upgrade to the city's Eye on the Street program will cut down on time needed to review video footage captured by the cameras.

Thunder Bay police currently training officers, will run pilot project in coming months

A computer with an automatic facial recognition system is shown in this 2017 Associated Press file photo. The new cameras purchased by Thunder Bay Police have the ability to use facial recognition, but a spokesperson says that would be used on a limited basis. (Markus Schreiber/The Associated Press)

Thunder Bay police say a new software upgrade to the city's Eye on the Street program will cut down on time needed to review video footage captured by the cameras.

The software, called BriefCam, makes use of artificial intelligence to make searching for specific footage easier and faster, said Chris Adams, the police service's director of communications and technology.

Adams said searching video captured by the Eye on the Street cameras is very labour-intensive.

"Say you have four cameras that you're searching, looking for a suspect, or you could be looking for a missing person," he said. "You could be looking for a vehicle that was involved in either a crime, an abduction, or something more generalized."

"To be able to search that, it required actually sitting an officer in front of a screen doing a playback for that, almost real time. Which is extremely inefficient, takes a long time."

The new software allows police to upload video to a server, and quickly search for a specific thing, such as an individual wearing a specific colour of clothing.

"What it does not eliminate, which is very important for everyone to realize, is the human connection in this," Adams said. "It still takes an investigator to use the tool to then go back to the original video, where those segments were, where an individual, a vehicle or whatever has been discovered, to actually then re-examine the original video to see if there is totality."

"In other words, look at the entire segment of that video that was recorded in there in that time frame."

Privacy is a factor in the use of the software, Adams said. The city's police services board is currently working on a policy that will govern the use of the new software, a draft of which is due back at a future board meeting.

Adams said the software is capable of facial recognition, but that feature would be "very, very limited" in how it's used in order to prevent errors, and potential misuse.

"In an instance where you had a missing person, and you knew that it was an extremely serious incident where there's extreme high risk, then you should be able to theoretically then use that AI feature of facial recognition to hopefully help you find that person," Adams said. "But having said that, that's why we're working very closely with the police services board to ensure that they have a governance model in place that basically is the check and balance to us."

Training underway

Adams said the BriefCam software has been purchased, but it is not yet in active use. Currently, officers are being trained, and a pilot program is expected to take place over the coming months to help police identify limitations and concerns with the software.

"We're not just going to flick the software on and just arbitrarily start using it," he said. "We really want to make sure that our investigators that are attached to it have a very deep understanding for how it works, what its limitations are, and then how effective it can be."

"And that's all done in conjunction with the governance model that we've asked the board to develop to make sure that we are doing it the right way."

The police service also contacted the Office of the Information and Privacy Commissioner of Ontario about using the software, the office confirmed.

"We are pleased that the Police Service has taken notice of our initial guidance and has committed to engage further our office as it takes steps to help build the necessary governance framework prior to program roll-out," the office stated in an email to CBC News. "We understand that the Police Service intends to consult with members of the public, as well as the Thunder Bay Police Services Board Administrator and key stakeholders."

"Such commitments are particularly important given that video surveillance cameras, facial recognition, and artificial intelligence technologies can have a profound impact on the privacy and human rights of Ontarians."

Video requests rising

Matthew Pearson, the city's manager of central support, which oversees the Eye on the Street program, said requests for video from city police are increasing.

"In the month of May, I think we're over 12 requests," he said. "Last year, May, there might have been one or two requests.

"I think the totals for last year were under 10 [requests] total, for the year," Pearson said. "Usage is trending up, for sure."

The Eye on the Street program recently saw a major upgrade of its camera equipment. There are currently 19 operating cameras in the system — another one is being repaired, and will be online soon. A list of camera locations is available on the city's website.

Pearson said privacy is front-of-mind for him, as well, when it comes to the Eye on the Street program.

"I am probably very similar to the average person in Thunder Bay walking the streets, about privacy concerns," he said. "I value individuals' privacy, all of our privacy."

However, Pearson noted the new software isn't creating any further privacy concerns, as the footage was already available.

"This is a speed tool," he said. "So then I revert back the camera system — is the camera system the right way to monitor?"

"I think the data will will support that it probably is," Pearson said. "It will produce results."