YOLOv3-based synchronous end-to-end license plate positioning and recognition method

Abstract

A YOLOv3-based synchronous end-to-end license plate positioning and recognition method comprises the steps of obtaining a license plate external rectangular frame and license plate number information,making a data set label file, and dividing a data set into a training data set, a verification data set and a test data set; training a license plate detection and recognition integrated network LPYolo based on YOLOv3, so that the LPYolo recognizes various types of license plates while detecting the license plates, augments a training data set, and puts the augmented training data set into the network for training; and positioning and identifying the license plate on the test data set by using the trained network model, and then evaluating the model. According to the invention, license platesof various types and different lengths can be classified and identified. Meanwhile, the network LPYolo proposed by the method can quickly and accurately detect and identify the license plate in various unconstrained environments.

Description

technical field [0001] The invention relates to the technical field of automatic license plate recognition, in particular to a YOLOv3-based synchronous end-to-end license plate location and recognition method. Background technique [0002] Automatic license plate recognition system is widely used in the field of intelligent transportation, such as vehicle violation monitoring, highway intersection charges, parking lot charges, etc. In road traffic management, the license plate is the only basis for obtaining vehicle owner information, so it is of great significance to accurately and effectively recognize the license plate information during the driving process of the vehicle. Existing license plate recognition technology is usually divided into two parts: license plate detection and character recognition. The existing license plate detection and character recognition technologies can be divided into traditional methods and deep learning methods. [0003] Traditional licens...

AI Technical Summary

Problems solved by technology

At present, the existing license plate recognition methods usually include multiple steps such as license plate detection and license plate recognition. Recognition errors, so more processing steps will have a greater impact on the final result of recognition

In addition, the more processing steps, the higher the time cost of the system

[0009] The existing license plate recognition technology usually includes multiple steps such as license plate detection, character segmentation, and character recognition. Each processing step will produce different degrees of error, and the error in the previous step will be further amplified in the subsequent steps, which will eventually lead to license plate recognition. not effectively

License plate images in real scenes are usually affected by various factors such as shooting angles and lighting, resulting in problems such as large tilt angles of license plates, compressed license plate characters, and sticky license plate characters, which will greatly increase the difficulty of character segmentation

Also, processing with multiple steps increases computational cost

Images