On November 15, 2017, the Shenzhen ‘Traffic Brain’ project, a collaboration between Huawei and the Shenzhen Traffic Police Bureau, was announced at the Smart City Expo World Congress. The project stood out from more than 300 leading smart city projects from around the world and won the only Safe City Special Award at the global event.
This prestigious award is the result of the Shenzhen Traffic Police’s continuous innovation and proactive police-enterprise collaborative efforts. The ‘Traffic Brain’ relies on a large number of technologically advanced applications, including facial recognition for traffic violation detection, traffic signal timing optimization (TrafficGo), and secondary image-based traffic violation identification.
Shenzhen’s ‘Traffic Brain’ is just the latest example of the bureau’s long history with digitally-assisted law enforcement. Historically, the Shenzhen Traffic Police have been early adopters of new technologies:
- In 1997, an electronic police (e-Police) device was integrated, which marked a distinct shift from manual to intelligent traffic law enforcement.
- In 1998, a video-based law enforcement application was launched. This technology enabled teams to work together, instead of relying on individual police officers on the street.
- In 2001, a license plate recognition system was built. This helped contribute to their ongoing, comprehensive, multi-level traffic security protection system, which maintained the security, order, and flow of the city’s traffic.
- In 2017, they collaborated with Huawei to create an Artificial Intelligence (AI)-based traffic management solution called TrafficGo for automatic signal timing optimization. This has redefined the way traffic is managed, shifting the focus towards a more-efficient vehicle flow.
- In 2018, they officially implemented facial recognition-based law enforcement, eliminating the need to detain and interrogate drivers and pedestrians.
These technological advancements are the direct result of the innovative spirit of the Shenzhen Traffic Police, combined with the continuous efforts of platform and application vendors.
ICT Enhances Three Levels of Intelligence
When the Shenzhen Traffic Police started working with Huawei and other vendors (including Intellifusion, SEEMMO, and Harzone) in 2017, layer decoupling and open innovation were the guiding principles of their collaboration. The solution leveraged video cloud, big data, and AI to enhance three levels of intelligence — computing, perceptual, and cognitive — building a unified, open, and intelligent traffic management system.
- Computing intelligence:The dynamic resource pooling solution used in the Shenzhen ‘Traffic Brain’ is based on Huawei’s open Atlas intelligent computing platform. With this solution, the Shenzhen Traffic Police are able to pool resources, including Field Programmable Gate Arrays (FPGAs), Graphics Processing Units (GPUs) which are currently in the spotlight, and future-oriented Neural Processing Units (NPUs).
Huawei’s platform helps facilitate law enforcement with its AI-optimized system computing. The platform allows Shenzhen’s traffic management system to intelligently adjust its services according to the dynamic changes to pedestrian, vehicle, and logistics flows on the streets. Compared with traditional platforms that allocate resources based on individual experience, Huawei’s platform enables more-accurate computing and efficient resource allocation.
- Perceptual intelligence:Based on the Enterprise Intelligence (EI) deep learning platform and open algorithm warehouse, the Shenzhen ‘Traffic Brain’ supports a broad range of algorithms from multiple vendors, including those for traffic volume, incident, and violation information collection, as well as secondary image-based traffic violation identification.
The open algorithm warehouse increases perceptual intelligence. For example, the algorithm for collecting traffic violation information can identify objects violating the law, including motor vehicle types (buses, trucks, vehicles transporting hazardous chemicals, and small passenger cars), non-motor vehicle types (personal bicycles, shared bicycles, electric cars, and tricycles), and pedestrian types (students, commuters, delivery personnel, and elderly people).
- Cognitive intelligence:Upper-level applications enhance the already-powerful cognitive capabilities of the ‘Traffic Brain.’ By leveraging EI-based intelligent traffic management, big data, and special databases for typical types of personnel and vehicles, the ‘Traffic Brain’ platform performs data analytics, makes comparisons, and ultimately learns how to facilitate safer and more-efficient traffic flows. Cognitive intelligence enhances when the platform analyzes traffic accidents caused by impaired drivers, vehicles transporting hazardous chemicals, and vehicles driving at night. This information helps law enforcement better understand the risks posed by these specific types of situations and come up with special solutions, and possibly regulatory measures.
The continued success of the ‘Traffic Brain’ solution emerges from the collaboration between the Shenzhen Traffic Police, Huawei, and many other ecosystem partners, including China Electronics Technology, Harzone, Ping An Technology, Intellifusion, SEEMMO, SenseTime, 1000video, and the Shenzhen City Traffic Planning Design Research Center.
The joint innovation team formed by these partners proactively met evolving development requirements; in fact, deployment of 40 e-Police devices using the world’s first facial recognition-based law enforcement application throughout Shenzhen occurred within 40 days. These devices were connected to the Shenzhen ‘Traffic Brain’ through optical transport networks, and Huawei’s Atlas computing platform and AI platform algorithm warehouse were adopted to enable these devices to intelligently perceive traffic objects and detect traffic violations.
Joint Innovation Drives Advanced Development
In order to implement facial recognition-based law enforcement, the Shenzhen Traffic Police built a platform to integrate algorithms from many vendors based on Huawei’s AI platform; created traffic-oriented facial image databases based on Huawei’s big data platform; and integrated the platform and databases into its unified application portal. In this way, layer decoupling ensures the accuracy of facial recognition. The results of the facial recognition are then delivered to the Shenzhen Traffic Police through a unified interface. This transformation, from vehicle-based to person-based traffic violation management, represents the innovative development of this project.
Another illustrative example is the Jaywalking Regulation System. The Shenzhen Traffic Police discovered that traffic violations are common among delivery drivers in non-motor and non-standard electric vehicles, often resulting in serious accidents. In response to this situation, Intellifusion, a Shenzhen-based innovative application vendor, developed the Jaywalking Regulation System to help the Shenzhen Traffic Police better monitor, detect, and fine jaywalkers by using facial recognition technology. Ultimately, the system also reduces accidents and prevents repeat offenses.
This system proved to be very effective. The first case involved a courier who ran a red light in the city’s Futian district while driving an electric vehicle to deliver goods. The facial-recognition e-Police device installed at the intersection captured a photo of the courier on the spot, which was sent to the backend system for analysis. The system scanned the database and identified the courier in seconds. The Shenzhen Traffic Police then sent the courier a fine in accordance with the city’s traffic regulations.
With the development and widespread use of AI technology in the transportation field, the Shenzhen ‘Traffic Brain’ will continue to evolve as an intelligent traffic management system. With an open, collaborative, and shared platform, Huawei, as well as many algorithm and application vendors, will continue to work with the Shenzhen Traffic Police to explore new concepts, models, and methods to safely and intelligently manage city traffic.