Putting the Brakes on Traffic Violations with AI
Editor’s Note: insight.tech stands in support of ending acts of racism, inequity, and social injustice. We do not tolerate our sponsors’ products being used to violate human rights, including but not limited to the abuse of visualization technologies by governments. Products, technologies, and solutions are featured on insight.tech under the presumption of responsible and ethical use of artificial intelligence and computer vision tools, technologies, and methods.
Nobody likes being stuck in traffic. But urban areas are overcrowded and suburban commutes are way too long. Plus, vehicles come in every shape and size—from tiny scooters to huge trucks—creating a mishmash of hazardous traffic.
Population expansion is only increasing the problem. Cities hold more than half the world's people and this trend continues to grow 1.8 percent annually.
One example is Bangalore—the Silicon Valley of India—notorious for its bad traffic. In 2017 the city had an estimated population of 12.34 million, up from 8.5 million in 2011.The average citizen sits in traffic for more than 240 hours a year. And management of all the vehicle movement costs the city 65 billion rupees, or $950 million annually.
Human resources alone are no longer sufficient to handle the volume and variety of vehicles traveling through today’s congested cities. To address these challenges, new technologies are emerging—enabling smarter traffic control systems.
The combination of smart video, artificial intelligence, and advanced hardware can monitor traffic and automate enforcement. These systems are becoming a vital tool for municipal governments to automate traffic control, detect violations, and even dispense tickets.
But many traditional solutions lack the flexibility to recognize, analyze, and learn the huge variety of vehicle types, license formats, ever-changing traffic patterns, and street layouts. Many cities have proprietary control systems that simply can’t keep up with the volume and variety of traffic.
While mentioning challenges with capturing correct license plate recognition, Avinash Trivedi, VP of Business Development at Videonetics, offered an example: “Some traffic management vendors try to fit the data, which they see from the camera, to a municipal standard plate. But when license size, lettering, or layout differ even slightly from the standards, analysis software can deliver faulty results.”
Videonetics Technology Private Limited developed its Intelligent Traffic Management System (ITMS) to address these problems. It’s a highly adaptable solution thatintegrates video cameras, sensors, and control systems.
Traffic Control Driven by AI and Deep Learning
Videonetics ITMS offers a unified architecture that enables traffic monitoring, control, and citation management. Its AI and deep-learning framework leverage extensive data from deployments in more than 100 cities across India. More than 7,500 traffic lanes are being monitored by its Intelligent Traffic Management System (ITMS).
Motorcycle helmet violations are a huge safety risk and serious problem in cities like Bhopal, India. When the company deployed its ITMS solution to monitor 14 intersections in the city, more than 650 people riding illegally without helmets were identified—in just the first six hours.
Even those riders wearing hats or other head coverings made to look like helmets were detected—which was not possible with the city’s previous traffic monitoring system. And now with ITMS software, the city can automatically send warnings or tickets to violators.
“Over a period of time, we have collected data of real-time situations from various cities and applied our new generation of AI and Deep Learning framework on these large datasets,” said Trivedi. “We have been able to extract and classify information based on local environment and train our AI engine to provide a high level of accuracy and object classification—with helmet detection being just one example.”
Delivering Distributed Traffic Analysis
The Videonetics solution acts as a true decision support system for traffic planners and traffic law enforcement agencies. It integrates Intelligent Video Management Software (IVMS) and Intelligent Video Analytics in a unified system architecture. This enables surveillance and traffic monitoring services to complement each other, addressing field issues with a holistic approach.
ITMS can work in both a central or distributed architecture. In a distributed system, multiple cameras are installed at each junction and connected to a local mini-server placed as shown in Figure 1.
The ANPR camera system automatically captures the license plates of any vehicle in the field of view (FOV) of a camera and stores them in a database, so that details of the vehicles are available at any later point in time along with related video footage. And other deployed cameras detect various violations, including jumping a red light, no helmet, triple riders, and others.
Once an event occurs, the associated information (camera name, junction name, time stamp, license plate image, license plate number, etc.) is sent to the central server and stored in a database. A report of all such events can be generated from the system. And video clips associated with the event can also be replayed on user request.
All the events detected and captured by the system are also locally stored at the junction server. If connectivity between the junction and the central control room is lost at any point in time, events and relevant video clips are not lost.
Videonetics has found Intel® technology to be an important enabler. “Intel provides a computing environment both in terms of CPU and GPU within a single silicon package,” saidTrivedi. “That's a huge advantage for us.”
Data Reporting and Management
ITMS gets raw data from visual analysis and then provides traffic management services to track, regulate, and analyze vehicle movement and enforce traffic rules for the safety of drivers and pedestrians.
System operators can search the event archives based on multiple criteria such as date, time, event type, license plate number, or any combination of these parameters. The search results can be exported as an Excel or PDF file. This enables reporting with evidence proof from multiple cameras.
Based on this data, the Videonetics Integrated e-Challan/e-Ticket Management Software (ICMS) can manage the citation process. An e-ticket can include visual evidence, fines, and payment deadline. The software can also provide the relevant information to courts in case a driver challenges the charge. Plus, ITMS also can connect with a government database to retrieve the identity information of a vehicle owner.
Videonetics solutions go well beyond red-light and no-helmet violations. The system can detect illegal smartphone usage, triple riding, speeding, suspected vehicles, sudden congestion, and much more.
The result is a robust intelligent traffic management system that can be tailored to work with the requirements of a given area and not a one-size-fits-all model. By using adaptable ITM systems like Videonetics, cities can improve their traffic management, protect citizens, and lower operating costs.