Mingjing Vehicle Big Data Analysis and Retrieval System is based on intelligent vehicle analysis and judgment products of video stream and picture stream. Based on advanced deep learning and big data mining technology in the world, it transforms massive unstructured data such as video and image into full information structured data and carries out deep data mining.
The product breaks the limitation of traditional vehicle detection and recognition which is limited to license plate number. It can automatically identify more than ten feature information of vehicles. It upgrades the traditional "license plate search" mode to "vehicle search" mode, and solves the problems of using suits, unlicensed, or occluding and defacing license plates, which widely exist in vehicle-related cases at this stage. The problems mentioned above provide a perfect solution for the control of vehicle violation cases.
The system has powerful functions and excellent performance, including five functional systems: vehicle information structure, vehicle intelligent retrieval, vehicle detection and control, vehicle data analysis and judgment, and vehicle search. It can support online real-time and offline analysis.
Product function
Structured Subsystem of Vehicle Information
Unstructured information such as pictures and videos can be recognized as structured information, which includes more than ten kinds of feature information such as vehicle number plate, body color, vehicle brand, vehicle type and vehicle feature.
Vehicle Big Data
Intelligent Vehicle Information Retrieval Subsystem
The system supports multi-condition combination intelligent retrieval according to body color, license plate number, vehicle brand, vehicle system, vehicle year, annual inspection mark, paper towel box, pendant, seat belt, sunshade, co-driver and so on.
Vehicle Large Data Analysis and Retrieval System
The Research and Judgment Subsystem of Vehicle Big Data Technology and Warfare Method
The vehicle technology and tactics research and judgment subsystem is based on the traffic data of the jack and the electric police. It provides all kinds of research and judgment tools for the characteristic information and driving behavior of the massive vehicle pictures. It helps the case handlers to identify the effective clue information of the case and lock the potential suspected vehicles.
The system provides abundant tactical and tactical functions. The main supporting techniques of large data analysis and judgment include:
Real-time alarm of license plate cars, information query of first-time vehicles entering the city, information query of frequent passing vehicles and information query of peer vehicles;
The functions of multi-vehicle inquiry, similar license plate concatenation, hidden vehicle mining, multi-point collision analysis, vehicle landing point analysis, etc.
For the police officers, the study document "Principles and Application Thoughts of Vehicle Techniques and Warfare" is provided.
Research and Judgment System for Vehicle Large Data Analysis
The Subsystem of Car Search Based on Maps
When suspected vehicles can not be retrieved by semantic search, they can be retrieved by directly inputting the relevant pictures of suspected vehicles as the retrieval condition, and the way of image search can be used to achieve the retrieval of vehicles, providing the maximum high-tech means to identify suspected vehicles.
The Vehicle Detection and Distribution Control Subsystem
The system can set up vehicle control rules, including monitoring points, features, license plate number, license plate type, body color, vehicle type, vehicle brand, vehicle system, vehicle payment and other conditions, find vehicles that meet the control rules, give an alarm immediately, and respond quickly.
Vehicle Big Data Detection and Control
System characteristics
Single server can handle millions of pictures
The top matching server can process more than 15 pictures per second, and can process more than 1 million pictures per day on average.
Tens of millions of data seconds quick check
Through the characteristics of body color, vehicle number plate, vehicle brand, vehicle type and vehicle features, tens of millions of data second query time can quickly lock the target vehicle.
High recognition rate of vehicle type
The recognition rate of vehicle type is over 98.6%.
Generation of multiple statistical reports
The system can analyze large data and generate statistical reports according to user's needs. Provide data support for front-line police officers and provide research and judgment basis for leadership and command.
Seamless docking with existing systems
The system provides a complete interface scheme, without revamping the existing interface system, supports rapid development integration, and achieves the application of rapid vehicle real-time investigation.