Vehicle recognition technology based on deep learning

The advent of the era of artificial intelligence has greatly improved the related work efficiency based on video image analysis. The emergence of intelligent image analysis systems and intelligent data analysis decision-making systems is gradually realizing the replacement of some human work content. The current application of deep learning-based network technology has made a qualitative leap in vehicle recognition technology under video image analysis.

Image analysis vehicle recognition technology Convolutional Neural Network (CNN), which is commonly used today, is a feedforward neural network. Its artificial neurons can respond to a part of the surrounding cells in the coverage area, and it has excellent performance for large image processing. The CNN-based video structured algorithm is a technology for extracting video content information. By means of space-time segmentation, feature extraction, target recognition and other technical means, the text structured analysis of video image content is realized. From the perspective of data processing, video structured technology can transform surveillance video into semantic information that humans and machines can understand.

By means of structured analysis of video images, people, cars, and objects in the scene can be accurately detected and identified, and they can be separated from the environmental background. In this way, through the structured analysis of the camera shooting picture, the number of targets in the front scene, corresponding target position information, and other available data can be obtained.

Through artificial intelligence analysis, vehicles in the scene can be detected and feature extracted. Under normal light conditions and normal night light, our vehicle identification module can detect vehicle types, brand models, Accurate identification of vehicle license plate information, body color, window markers, illegal driving behaviors, etc. We have identified the vehicle's various features with an accuracy rate of over 95%. Among them, the vehicle type identification item can identify 21 types of vehicles, which is more detailed than the 9 types of vehicle types required by the national standard. In the vehicle brand and model identification, the front can identify more than 5,500 types and the rear can identify more than 3500 types. , Such as muck trucks, hazardous chemicals transportation vehicles, trailers, cranes, etc. for effective identification. At the same time, it supports the vehicle's arbitrary angle recognition, and supports vehicle target re-authentication and trajectory tracking.

Main advantages of license plate recognition parking management system:

1. Small cost-no need to increase the cost of consumables of the traditional credit card system, the minimum human resources cost

2. Automatic recognition of license plates: Automatically identify vehicles entering and leaving, and create an intelligent vehicle management system

3. Automatic control-the system does not require human intervention, intelligent management and control, reducing human loopholes

4. Flexible management-detailed records of vehicle entry and exit data, flexible statistics of various data reports

5. Video sharing-The license plate recognition HD camera's video source is shared with the monitoring center, saving extra investment

　　 License plate recognition parking management system function

1.Entry fast without stopping

There is no need to issue a card invoice at the entrance, and no card inspection and ticket inspection at the exit. And can achieve unmanned entrance, barrier-free access. You can save the cost of cards or tickets, and correspondingly reduce personnel costs.

Automatic license plate recognition

The license plate recognition system uses the license plate number of the inbound and outbound vehicles as a voucher for vehicle management, automatically collects the front characteristic image of the inbound vehicle, automatically recognizes the license plate number and records the front panoramic image of the vehicle, as the original data for parking management and safety certification. And use the license plate number as the data identifier for information and data management.