Grasps increasingly complex, real-time urban conditions with AI, supercomputer technology
Fujitsu today announced development of solutions that make use of images from surveillance cameras deployed throughout a city or inside a facility, using high-speed image processing technology cultivated with its Human Centric AI Zinrai artificial intelligence and supercomputer technologies. FUJITSU Technical Computing Solution GREENAGES Citywide Surveillance software, which enables a real-time understanding of urban conditions, and FUJITSU Technical Computing Solution GREENAGES Parking Analysis software for intelligent parking management, will be available for sale in Japan from today.
Using image analysis that employs advanced AI technology, Citywide Surveillance automatically extracts a variety of data from surveillance camera images, such as the type or make of vehicles, or the type of clothing someone is wearing, giving users a real-time understanding of movements throughout an urban environment. For customers with monitoring needs beyond vehicles or people, Fujitsu’s AI experts will provide AI learning services with expanded image analysis capabilities, built on highly accurate recognition models.
Parking Analysis supports improved management efficiency of parking lots or garages by ascertaining which parking spots are open or monitoring no-parking zones in real time.
By watching out for entire cities, Fujitsu seeks to create a human-centric intelligent society, where people can live with greater prosperity and security.
With the rapid rise of urban populations and the increasing complexity of economic activity resulting from the greater movement of people and things, protecting people from crime, natural disasters, and accidents, and creating urban environments where they can live with peace of mind, are becoming urgent social issues around the world. To address such issues, the use of surveillance cameras is spreading, but to grasp conditions over the wide area of an entire city, several thousand cameras may be required, and there is a limit to what can be ascertained just using visual observations by people.
In 2015, Fujitsu conducted a field trial of the new Parking Analysis solution in Dubai, United Arab Emirates, where it was utilized with monitoring cameras positioned around outdoor and basement parking lots, and other areas, to verify the accuracy of its grasp of parking conditions, number cars, and other information. In both the Parking Analysis and Citywide Surveillance solutions, Fujitsu employs Zinrai’s cutting-edge AI image analysis technology developed by Fujitsu Laboratories Ltd. and Fujitsu Research and Development Center Co., Ltd. in China. This technology makes it possible to grasp the movement of people and vehicles throughout a city, thereby helping to prevent crime and optimize urban lifestyles.
Features of the Solutions
Using sophisticated AI technology in which a recognition model is built based on learning from large volumes of imaging data, and with image recognition using deep learning based on that model, Citywide Surveillance can ascertain such characteristics as the make of a vehicle or the type of clothing someone is wearing. Using distributed, high-speed image processing technology that Fujitsu refined with supercomputers, this data can be extracted from surveillance camera images in real time with high precision, enabling users to grasp the movements of vehicles or people throughout a city. This solution can be widely used in facilities, such as airports, train stations, major roads, industrial parks, shopping malls, and parking garages, or in public spaces, such as in urban areas, to grasp overall conditions, prevent crime, or for other applications.
2. Major features
- Vehicle identification: Ascertains the type, color, make, and model of vehicles.
– License plate identification: Reads the license plate numbers of vehicles.
– Human identification: Ascertains color and clothing type worn on the upper and lower body.
– Counting vehicles or people: Counts vehicles by type or people, recording the time of detection.
– Facial recognition: Automatically tracks and searches a person’s face, and detects faces that have been registered in advance.
Figure 1. Identification of vehicles, people, and faces
3. AI Learning Service (option)
In the Citywide Surveillance solution, aside from the built-in image recognition model for vehicles and people, if customers want to add their own customized recognition targets, Fujitsu’s AI experts will build a sophisticated recognition model for which functions can be flexibly expanded.
With Parking Analysis, the status of up to 100 parking spots per surveillance camera can be ascertained in real time. In addition, for cases where wide-area monitoring cannot be performed, such as an indoor parking garage, images can be used to count vehicles as they enter and exit, and the number of open spots in each zone can be calculated.
As a result, customers who operate parking lots can use this service for a variety of objectives, including for such improved services as enabling smartphones to determine availability of open parking spaces, enhanced security by monitoring no-parking zones or cars that are parked for long periods of time, and smarter marketing, such as optimizing earnings by setting parking rates in accordance with the number of open parking spots.
By combining this solution with Citywide Surveillance, even more sophisticated parking management can be achieved, such as managing more detailed vehicle data, including types of cars and license plate numbers, or monitoring vehicle break-ins.
2. Major features
- Monitoring of parking spots and zones: Immediately ascertains the number of open spots, or determines the number of vehicles in a given zone.
- Anomaly detection: Automatically detects parking in no-parking zones or parking that straddles two different spots.
- Time series analysis: Uses past data to display changes in usage status over time (daily, weekly).
Figure 2. Image of the administrator’s screen
Pricing and Availability