Invision AI successfully completes high performance portable roadside Vehicle Occupancy Detection (VOD) field tests with over 97.5% precision

Toronto, March 30, 2021 – Invision AI, a leading provider of video-based AI solutions for transit and public safety applications, today announced the results of testing the Invision Road-Side Unit (iRSU) at three sites managed by global toll road developer and operator Transurban.

The iRSU was deployed in three different sites in the Greater Washington Area and achieved a precision of 97.5% in under two hours. The low-occupancy vehicle miss rate was 11.9%. Operated in “enhanced mode,” the iRSU was proven to be capable of delivering a 99.0% precision at the cost of a slightly larger miss rate of 15.9% with the same 2-hour set-up time.

“We set out two years ago to disrupt the vehicle occupancy detection market with a solution that moves away from manual reviews to full-automation with precisions reaching or exceeding 99%. We wanted to offer a cost-effective solution that requires no site-specific tuning and can be moved from site to site if desired” says Karim Ali, CEO at Invision AI.  

“We are excited to see this mobile Vehicle Occupancy Detection project be demonstrated successfully. Transurban is always looking for ways to work with companies like Invision AI who are putting technology to work to make our roads safer and more efficient,” said Lev Pinelis, Director of Innovation at Transurban North America.

The iRSU supported multiple lane monitoring, quick operational readiness after relocation and expected performance. Transurban’s concept of a mobile system seeks to provide a more efficient enforcement and compliance system to ensure fair travel for customers on its Express Lanes network.

Invision is set to ship its iRSUs over the next year to multiple road operators across the globe and is in advanced partnership discussions with several multinational technology conglomerates. The iRSU may be installed as a fixed system, where it achieves 99.1% precision and a 2.4% miss rate, or can be kept on its portable platform for rapid deployment in different sites. The iRSU can interface to different gantry systems as well as license plate recognition systems to give a full report of each vehicle, its license plate and registration information, its number of occupants in different rows, and confidences associated with all these predictions.

To address security and privacy, no vehicle or occupant data is retained or transmitted over the air. All computations are efficiently performed on the edge and the client can choose to keep only suspected violator information encrypted on the edge or securely upload these encrypted data to its own cloud infrastructure.