It’s no secret that it can be complicated to find legal parking on Montreal streets. The rules are endless: road cleaning schedules, permit-only stretches and traffic lanes, to name a few.
Standing in front of a pole with multiple parking restriction signs attached to it in Old Montreal, one driver marveled at how difficult it is to understand and quickly summed up the headaches behind finding a safe spot.
“The stars must align,” he said. “Jupiter and Saturn must align. It’s a little weird.”
But what if there was an app for that?
A Quebec company launched a new AI-driven application in May to help drivers decipher confusing parking signs and, ultimately, avoid steep fines and having their vehicles towed away.
Underlabs came up with the idea of Parky.ai after one of its members, its chief technology officer, was confused by a sign and came back to find a ticket on the windshield.
“We thought that artificial intelligence could solve this problem,” said Andrei Chira, senior software engineer at Underlabs. “So we decided to create the most simple app where you just take a picture of all the parking signs and boom — it tells you if you’re allowed to park there using artificial intelligence.”
All drivers have to do is take a photo of the parking restrictions in the app. Parky.ai then processes the photo and gives a yes or no answer on whether you can safely park there or not.
How does it learn all that? Underlabs developers took more than 1,000 pictures of different signs to teach the app how to decode the city’s plethora of parking regulations.
“Our goal was to provide an answer faster than humans would be able to read and interpret the signs for themselves,” Chira said.
While the initiative is meant to take the stress out of driving in Montreal, Parky.ai isn’t perfect. The AI-driven app is still learning and motorists who use it also help hones its skills.
At the moment, the company says the current accuracy for the app is 83 per cent for single signs and 74 per cent for multiple signs. If the user notices a mistake, they are encouraged to flag it.
“We suggest people report inaccuracies in the app. There is a button for that because that will allow us to train the model even further,” Chira said.
Parky.ai is focused on Montreal for now as it’s still new, but Chira says Underlabs could take their idea to other provinces down the road.
— with files from Global News’ Mike Armstrong