bg line bg line

HiSilicon Smart Vision, Making Cities Safer and More Secure


[Shanghai, China, Jan 14, 2021] Public safety is the bedrock of civilization, and a basic right in the modern world. As cities have grown in size and scale, protecting vulnerable populations has also become a major challenge. It is projected that by 2030, 5.1 billion people will live in cities, and that a staggering 55% of the world's population will live in medium-to-large cities. The number of mega-cities, those with more than 10 million people, will soon soar to 41.

City management

Traditional city management has been hindered by a lack of data, inefficient operations, and inadequate security capabilities. Burgeoning technologies like artificial intelligence (AI) and the Internet of Things (IoT) have helped spur the creation of smart cities, making data ubiquitous thanks to intelligent perception capabilities. AI has transformed videos into structured information, rather than mere streams of code, extracting information such as license plate numbers, vehicle colors and models, and thereby multiplying the value of video data. Technological development has led to efficient public safety services. As new smart devices advancements are unveiled in the near future, these trends will only accelerate.

Smart Vision, Safer Cities

Broader Application of Smart Vision

• Subdividing scenarios: Commercial applications for security solutions are abundant, from access control and parking lots, to turnstiles and geofences. These are applied across a wide range of contexts, on a city-wide, community-wide, or campus-wide scale; each of these environments represents a flourishing market in its own right. The broad deployment of intelligent solutions will offer unprecedented convenience for residents, and make the long-held smart city ideal a viable reality. With HD video shooting, transmission, storage, and presentation, as well as AI-assisted data analysis, and decision-making, resources and services can be allocated in an optimal manner, to provide real-time protection for residents, and proactive alerts. Seamless clock-in/out enabled by facial recognition and encoded privacy protection also promises to make work life newly convenient. NB-IoT and license plate recognition will enable smart parking and automatic payment solutions, so that parking efficiency can be enhanced.

• AI: Thanks to its support for ultra-high bandwidth wireless transmission, intelligent IP camera, and digital video recorder (DVR), smart vision can be flexibly deployed for mobile surveillance and law enforcement purposes, on shuttle buses, tour buses, and dangerous cargo carriers, as well as in temporary locations such as construction sites and concerts. This will also reduce the reliance on fixed network and leased optical fibers, thereby slashing O&M costs.

How Smart Vision Enlightens Devices

• All-weather perception: Images can now be captured at all hours, and in poor lighting conditions; objects in images are detected, extracted and analyzed on an intelligent basis.
• 8K UHD: UHD videos come with abundant details, which means that it offers more analytical value for deep learning algorithms.
• Smart video coding: The pooling of AI and smart video coding (SVC) has allowed for the allocation of bit rates for essential content, which in turn reduces storage and transmission–related costs.
• Multi-source perception: The in-depth unification of images and perceptional data, including infrared images and sound, has allowed for seamless efficient IoT perception devices, while boosting recognition accuracy and maximizing the value of data.

Smart city

Cities require HD video surveillance on some level, in order to become optimally safe, smart, and conducive to the modern world. Urban surveillance has shifted to a new paradigm, from mere observation to smart perception.

HiSilicon AI SoCs

HiSilicon AI SoCs come with multi-channel any view stitching (AVS) on devices, and powerful 4 TOPS/2 TOPS computing, supporting AR smart city command systems that are able to fully harness augmented reality (AR) and 3D positioning, alongside AI pattern recognition and event detection. They make use of AR high-point panoramic cameras and low-point cameras in the field of view (FOV), to implement a comprehensive 3D monitoring system that observes both the whole and the part, with support for picture-in-picture displays for low-point camera videos, on top of searching, positioning, querying, and alarm capabilities. This high level of cohesion results in surveillance systems that are optimized across the board.

Every day takes a step closer to full-scale digitalization, and the next-level smart solutions that promise to enrich city life. HiSilicon AI has worked hand-in-hand with clients and partners throughout the ecosystem to make these solutions available ahead of schedule, so that today's cities enjoy tomorrow's capabilities.