Skip to content
Bike Lane Violation Detection with Mask R-CNN — AI & Machine Learning project by YenkoDev

Bike Lane Violation Detection with Mask R-CNN

Software as a Service

No

Category

AI & Machine Learning

Clients

Confidential

Techstack

PythonDetectron2Mask R-CNNOpenCV

Purpose

Cars and motorcycles drifting into bike lanes put cyclists at real risk, but catching violations by hand means someone watching hours of CCTV footage. The project set out to give traffic monitors an automated eye on the road that could flag encroachment on its own, reliably enough to be useful.

Description

We built a computer-vision system that watches CCTV footage, outlines each vehicle on the road with pixel-level precision, and measures how close it comes to a bike lane an operator marks directly on the frame. It tells cars, motorcycles, buses, and trucks apart, ignores bicycles, and flags any vehicle crossing into the lane. Teaching a model to separate and track individual vehicles in live traffic is a demanding vision problem, the kind we take on and ship.