Abstract
Roadside construction is a common sighting in a road environment. Construction workers put up sign boards for incoming road traffic to warn them of an ongoing construction work. But despite these warning boards, workers are in continuous danger of a speeding vehicle or an unaware driver crashing into them. To improve traffic safety, this project will survey exiting safety technologies and propose a new safety warning system that implements an image processing-based construction site detection using a dedicated camera and a computer. Although software frameworks that detect objects (cars, bicycles, animals etc.,) exist, these solutions are computationally expensive. The algorithm proposed in this project serves as a low complexity alternative that detects construction sites and provide early warning to drivers. Specifically, the proposed algorithm will detect sign boards pertaining to construction sites from real-time video feed. Upon detection of the sign boards and depending on the proximity of the vehicle, the system will generate a warning which will request the driver to slow down the vehicle to minimize the risk of accidents.