How to Improve Traffic Safety with AI Analytics
Pedestrian and vehicular traffic is a huge part of daily life in every city, impacting everything from quality of life to the local economy. It is essential that all transportation and traffic systems always prioritize safely and efficiency. Unfortunately, the 24/7 management of traffic systems poses a significant challenge for elected officials, city managers, and transit personnel alike.
To be successful, city personnel must be constantly informed of emerging situations, equipped with the right tools to prevent and respond to such events, and able to provide critical services where needed. But using manual methods to accomplish these tasks can be both labor-intensive and inefficient. For example, deploying crossing guards to facilitate safe pedestrian travel requires costly labor resources that cannot be relied upon through all hours of the day and night. Similarly, security personnel cannot be expected to scan all city surveillance cameras for traffic events and vehicles of interest efficiently. As a result, municipalities are wasting valuable time, money, and manpower by manually managing their traffic systems.
ISS’ suite of intelligent analytics powered by Intel are proven to resolve new and emerging traffic challenges by empowering existing video surveillance systems with the ability to autonomously monitor traffic conditions and to detect potentially harmful events. Analytics such as traffic violation detection and license plate recognition automate many otherwise manual processes and allow cities to distribute traffic resources including personnel and funds more efficiently.
Here are examples of how powerful analytics can work together to create Intelligent Traffic Systems capable of keeping cities safer with greater efficiency.
License Plate Recognition
Tracking and identifying vehicles of all types has been a longstanding challenge across the transportation industry. Doing so using standard surveillance methods is virtually impossible given the volume and speeds at which most vehicles travel. License plate recognition, or LPR, is a key tool in the ongoing effort to automate and enhance public safety for a variety of applications. And while many LPR solutions can be inaccurate (such as mistaking an “8” for a “B”) in instances of high speed or inclement weather, SecurOS™ Auto License Plate Recognition (LPR/ANPR) overcomes this challenge with the ability to accurately capture license plates at speeds up to 155 mph and in all weather conditions, including fog, rain, and snow.
The SecurOS™ Auto LPR solution was developed using ISS’ advanced artificial neural network technology in conjunction with Intel OpenVINO™ technology. The Intel distribution of the OpenVINO toolkit makes the most of Intel AI hardware, the latest neural network accelerator chips, and embedded computers and edge devices to maximize AI analytics performance. With over 20 years of proven results, SecurOS™ Auto LPR does not require the use of sensors and is hardware-agnostic. This means no specialized cameras or hardware are required for use, keeping deployment costs low. Plates of interest can be added to local watchlists and seamlessly connected with other external databases for quick search features. The system can set custom alarms when matches occur, ideal for law enforcement or border applications. For advanced applications, cities can employ the ‘Convoy Search’ feature that allows users to search for vehicle co-occurrences, detecting one or more unknown vehicles traveling in convoy with known target vehicles.
Traffic Violations Detection
There are almost an unlimited number of traffic violations committed by drivers every day. From running a red light to those all-too-common speeding infractions, these traffic violations have a serious impact on public safety while also placing strain on public resources. Law enforcement officers are trained to spot such infractions and issue citations; however, it is impractical for officers to dedicate 100% of their time toward detecting traffic violations and issuing tickets.
SecurOS™ Crossroad, powered by Intel, was developed to automatically detect traffic violations including red light violations, speeding, stop-sign violations, wrong-way traffic, road signs non-compliance, pedestrian right-of-way infractions, and other violations with high accuracy. SecurOS™ Crossroad is powered by SecurOS™ Auto ANPR which ties detected violations to the license plate number of the vehicle. In this way, the evidence gathered from SecurOS™ Crossroad minimizes human efforts required for traffic law enforcement activities, such as ticket issuing and vehicle tracking. The best part is the software works 24/7/365, so no violations go unnoticed or undocumented and officer resources can be redistributed to mission-critical tasks.
Pedestrian Crossing Safety
Pedestrian safety is an often-overlooked aspect of traffic and transportation management, but the fact is, pedestrians accounted for about 17% of all motor vehicle crash fatalities in 2020. Whether it be distracted driving, poorly lit crosswalks, driver fatigue, or a culmination of other factors, pedestrians are at risk of being a victim of a motor vehicle accident often when they least expect it. Pedestrians can also be a cause of accidents, as drivers may quickly break or swerve to avoid hitting a pedestrian. One solution may be to install more traffic lights; however, this can lead to unmanageable traffic levels or gridlock. Another solution is to install more road signs or lighted displays alerting drivers to a pedestrian crosswalk, but these solutions are costly and takes a driver’s attention away from what matters – the pedestrian.
With the SecurOS™ Soffit from ISS, pedestrians are escorted across crosswalks by dynamic illumination to prevent traffic accidents. This hardware and video processing solution uses AI algorithms to detect pedestrians, accurately determine their location in each video frame, and track their trajectory according to changes in their movement. The result is a well-lit pedestrian which significantly increases the speed of drivers’ visual-motor response, so they can brake in time to avoid accidents. What sets the SecurOS™ Soffit apart is that it is the only solution on the market designed to operate automatically 24/7 by employing video analytics detectors based on artificial intelligence technologies using cameras built on an Intel platform. Furthermore, SecurOS™ Soffit may operate with SecurOS™ Crossroad and can ensure the detection of traffic violations, such as failure to yield for pedestrians.
Under Vehicle Surveillance
Airports, stadiums, and critical infrastructure sites are all potential targets of domestic and foreign terrorist attacks. As a result, the detection of explosives has emerged as a key risk mitigation element for cities and facilities of all sizes. The underside of a vehicle is largely known to be a key point in the inspection of vehicles to prevent such an attack, making undercarriage inspections an important risk reduction tool. However, using manual methods such as mirrors to conduct visual inspections of a vehicle’s undercarriage is both laborious and ineffective.
The SecurOS™ Under Vehicle Surveillance System (UVSS) employs both hardware and software to simplify the undercarriage inspection process. The system’s unique combination of imaging hardware and intelligent analytics, using artificial neural network technology from Intel, captures images of a vehicle’s entire underbody and analyzes the image data to detect anomalies, such as explosives or contraband that may be affixed to the vehicle. The captured image is then stored, along with the detected license plate gathered via SecurOS Auto, for future reference. Moreover, it is a cost-effective and convenient solution that allows for the for visual inspection of all types of vehicles without impeding traffic flow.
The AI-powered analytics discussed here are just the beginning of ISS’ Intelligent Traffic Systems solutions. We welcome the opportunity for our team of transportation technology experts to discuss your facility’s challenges and needs, and which combination of intelligent analytics are most appropriate for your specific application. We look forward to hearing from you. Contact us today!