With the acceleration of urbanization, the development of urban transportation is facing a severe situation and new
challenges. At present, relevant departments are promoting the integration of new technologies such as 5G, big data,
artificial intelligence, cloud computing, block-chain and super computing with the transportation industry, so as to enable
the development of transportation with data resources, we will accelerate the integrated development of transportation
infrastructure networks, transportation service networks, energy networks and information networks, and build an
advanced transportation information infrastructure. It is the trend of the development of the transportation industry to
construct the comprehensive transportation big data center system, deepen the development of the public service and the
E-GOVERNMENT, promote the integration of the urban transportation system, and improve the overall quality of the urban
Developing intelligent transportation is a task that can not be solved by a single industry. Each industry has a relatively
deep area of innovation, but when it comes to integrating transportation services as a whole, it's not enough to focus on
just one industry. At present, the challenges of intelligent transportation in China are mainly as follows: first, data
perception, although there is a large amount of data, there are still a lot of data that are unknown or difficult to obtain;
second, decision support, we need to consider how the collected data can really create value and provide technical
support for emergency decision-making. The third is integrated management and control, that is, for the urban governance, there is no integrated management and control network.
Overview of AI video big data technology
AI Video big data can be understood as video acquisition is the foundation, artificial intelligence (deep learning) as the
carrier, big data application is the soul. Real-time monitoring of traffic conditions, traffic accidents, meteorological
conditions and traffic environment of scenes such as highways, tunnels, bridges, airstrips, railway tracks and customs ports
is carried out by means of 360 degree, Panorama, 180 degree Panorama, linkage of panoramic details and 4K high-definition video, relying on advanced artificial intelligence technologies such as vehicle detection, face recognition, image recognition continuation, and computer information processing, to obtain information about traffic conditions, and to process and analyze large data based on the information collected, to control the traffic, to assist the traffic management personnel for traffic command and dispatch, curb traffic violations, maintain traffic order, but also to assist public officers for control, Criminal Investigation and so on.
In domestic and international important occasions, many national leaders to AI video big data significance of the
exposition, the world today who have more AI video big data, who can be in the information of the rapid response capacity than other countries. All countries want to obtain as early as possible accurate access to and application of information on the movement of people in cities, traffic patterns, rapid response plans for emergency rescue, and the protection and exploitation of natural resources in forests, grasslands and oceans, to better provide video big data for economic development and social management, but also provide important digital information for National Defense Security and public security of the international community. The following highlights AI video big data in the big transport industry applications.
1. Be fully aware
For large scenes such as airport terminals, flight zones, railway departure floors, squares, traffic junctions, ports and
harbors, multiple cameras are needed for full video coverage, dahua uses 360 ° panoramic equipment to realize one equipment to complete the panoramic scene monitoring. On the basis of panoramic monitoring, we can use the
high-power ball machine to track the details, realize the linkage of panoramic and detail tracking, and fully perceive the big scene and detail. At the same time, the back-end of the server can also be used to splice the front-end of a number of devices to achieve a global perception of the screen, the ball machine can be linked to the details of tracking.
Figure 1 above is a panoramic video surveillance of the runway area and the far/near apron of an airport, while satisfying the effect of panoramic video surveillance in extreme weather conditions such as rain, snow, fog, lightning and night. It can display the high and low point video associated with the low position camera of the flight bridge, and the electronic magnification display in the panoramic video mode.
The range of traditional video detection and monitoring is 15-80 meters on highways and highways, but the effective
detection range of radar can reach 200 meters. By fusing radar data in video, the perceived range can be effectively
enhanced, provide richer data packets for higher level applications. DAHUA uses radar combined with video surveillance is equivalent to radar has a visual eye, radar detection function with the screen can be real-time monitoring, after the event to take back.
In the scene of road, subway station and so on, there is the problem that the traditional short-focus gun can see the short view but can not see the long view. The dynamic mosaics fusion technology of DaHua view can fuse two images of short and long focal length on the same frame in a binocular camera, realize the illegal action capture and license plate recognition, and realize the "clairvoyance" function, achieve"not farsighted, not short-sighted".
To solve the problem of video perception in each scene, Dahua ensures that the perceived data in each scene can be seen, seen and seen clearly through video mosaic, double pupil and double frame fusion techniques, through an intelligent algorithm.
2. Integrated Control
Based on the improvement of video perception and the harmonious distribution of algorithm and arithmetic power, the traffic data are cleaned and normalized, and the useless, error and redundant data are eliminated to ensure the quality of the data, the big data analysis model is constructed through the hierarchical integration. At the same time, it can form a set of scientific and reasonable integrated traffic model base by updating data input, model test and evaluation, and improve the whole integrated traffic safety monitoring system platform. Figure 2 below is a place to build a "car easy to manage" integrated control of the big data platform.
"vehicle easy management" based on Gis map show in the construction site information, static information, vehicle dynamic information, path information. For example, in the form of OD, we show the transportation situation of the spoil on that day: How many spoil trucks are transported from each construction site, how many spoil trucks are processed by each waste treatment site, and how many spoil trucks are transferred through the wharf. The off-line status data and alarm data of vehicles in the city on the same day are visually displayed, and the real-time position of the vehicles and the whole transportation process of the vehicles are monitored in real time, once found that there is a violation of the system to automatically alarm. The display of statistical analysis data in digital form can play an assistant role in decision-making and improve the efficiency of supervision.
In addition, focusing on the difficulties of street grass-roots governance, the main line is event closed-loop, to achieve a map management global object, a platform multi-collaborative, to meet the needs of grass-roots social governance in many scenarios. In the light of urban management pain points such as illegal parking, out-of-shop operations and itinerant vendors, changes were made to the previous manual discovery and disposal methods, and the front-end intelligent sensing alarm and automatic circulation of platforms were realized through "using machines to replace people" , thus reducing the work burden of the grassroots, improve management efficiency. For illegal parking, once there is a vehicle into the illegal parking area, the front-end automatic perception and linkage to the scene speaker remote call to remind the owner to leave in time.
3. Application trends
Big Data is a strategic resource for transportation power countries to transform into transportation power countries in the
next stage. The Ministry of Transport of the People's Republic of China Action Plan for promoting the development of
comprehensive big data in transport clearly states: "to build a technical model for comprehensive big data analysis, and to
study the establishment of a comprehensive and global big data analysis model with strong application value. ". In the
future, the AI video big data pool can be formed by integrating and aggregating AI video big data of multi-department,
multi-system, multi-level and multi-region, and the application value of big data can be deeply mined by using big data
model base and Algorithm base, presenting the whole comprehensive traffic situation, movement situation and
development level in the way of digitalization, visualization and Panorama, as a key data support for the real-time
monitoring and analysis of traffic managers, change from experience decision to data decision.
In addition, the combination of big data and Internet can construct all kinds of scene construction information service system based on traffic trip, integrate all system resources by using mobile terminal and Mobile Internet technology, construct the whole scene traffic trip ecological circle.
The construction of AI video big data in intelligent transportation needs to constantly improve the construction ideas, continuously promote value mining, and actively strengthen the open sharing of AI video big data, only then can provide the reference significance for the entire AI video big data center construction, the data governance and so on work, the ultimate goal uses for serves the society, serves the populace.