{"title":"基于ALPR系统的无约束场景下车牌识别","authors":"Zhiquan Jiao, Hongri Fan","doi":"10.1145/3366194.3366290","DOIUrl":null,"url":null,"abstract":"Most of the existing license plate recognition technologies mostly consider the frontal angle of view of the license plate images, and it is difficult to effectively solve the license plate recognition problem in complex application scenarios such as the distortion of the license plate area. Aiming at this problem, this paper does some studies based on the ALPR system in unconstrained scenarios. The vehicle detection network of the system is changed to YOLOv3 network due to its fast execution and high precision. The loss function in the original system is improved from the single function to the compound loss function. Part of the complex scene pictures which contain license plates from BDD100K dataset are selected as the test set, the initial ALPR system and the improved system's recognition effect on the test set are tested separately. The results showed that the accuracy of the original system in the selected test set for license plate recognition is 92.27%; After YOLOv3 is applied to the vehicle detection network and the loss function is improved, the recognition accuracy is 95.26%, which enhances the detection ability of small target vehicles while improving the recognition accuracy.","PeriodicalId":105852,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics, Intelligent Control and Artificial Intelligence","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2019-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"License Plate Recognition in Unconstrained Scenarios Based on ALPR System\",\"authors\":\"Zhiquan Jiao, Hongri Fan\",\"doi\":\"10.1145/3366194.3366290\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Most of the existing license plate recognition technologies mostly consider the frontal angle of view of the license plate images, and it is difficult to effectively solve the license plate recognition problem in complex application scenarios such as the distortion of the license plate area. Aiming at this problem, this paper does some studies based on the ALPR system in unconstrained scenarios. The vehicle detection network of the system is changed to YOLOv3 network due to its fast execution and high precision. The loss function in the original system is improved from the single function to the compound loss function. Part of the complex scene pictures which contain license plates from BDD100K dataset are selected as the test set, the initial ALPR system and the improved system's recognition effect on the test set are tested separately. The results showed that the accuracy of the original system in the selected test set for license plate recognition is 92.27%; After YOLOv3 is applied to the vehicle detection network and the loss function is improved, the recognition accuracy is 95.26%, which enhances the detection ability of small target vehicles while improving the recognition accuracy.\",\"PeriodicalId\":105852,\"journal\":{\"name\":\"Proceedings of the 2019 International Conference on Robotics, Intelligent Control and Artificial Intelligence\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-09-20\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Proceedings of the 2019 International Conference on Robotics, Intelligent Control and Artificial Intelligence\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1145/3366194.3366290\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 2019 International Conference on Robotics, Intelligent Control and Artificial Intelligence","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3366194.3366290","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
License Plate Recognition in Unconstrained Scenarios Based on ALPR System
Most of the existing license plate recognition technologies mostly consider the frontal angle of view of the license plate images, and it is difficult to effectively solve the license plate recognition problem in complex application scenarios such as the distortion of the license plate area. Aiming at this problem, this paper does some studies based on the ALPR system in unconstrained scenarios. The vehicle detection network of the system is changed to YOLOv3 network due to its fast execution and high precision. The loss function in the original system is improved from the single function to the compound loss function. Part of the complex scene pictures which contain license plates from BDD100K dataset are selected as the test set, the initial ALPR system and the improved system's recognition effect on the test set are tested separately. The results showed that the accuracy of the original system in the selected test set for license plate recognition is 92.27%; After YOLOv3 is applied to the vehicle detection network and the loss function is improved, the recognition accuracy is 95.26%, which enhances the detection ability of small target vehicles while improving the recognition accuracy.