With the increasing frequency of inland waterway shipping, effective management and monitoring of ships have become particularly important. As the only identification of a ship, the ship license plate is of great significance in the fields of lock management, inland waterway ship dispatching and port management. However, due to factors such as ship motion, water reflection, different sizes of ship plates and complex background, the detection and recognition of ship plates face many challenges. To this end, this paper proposes an end-to-end ship plate detection and recognition algorithm based on YOLOv8 and Improved Convolutional Recurrent Neural Network (YOLOv8-ICR). In the part of license plate detection, we use the YOLOv8 algorithm as the main body, combined with the CSPDarknet structure improvement, the introduction of the attention mechanism, the use of adaptive anchor boxes and hybrid loss functions, etc., to enhance the model's ability to quickly and accurately detect the license plate area in complex backgrounds. In the part of ship license plate recognition, we improved the CRNN algorithm. We used a deeper convolutional neural network structure to improve the feature extraction capability, and adopted a bidirectional long short-term memory network (BiLSTM) to capture the context information of the character sequence effectively and to improve the recognition accuracy further. Finally, the LeakyReLU activation function was used instead of the traditional ReLU activation function to solve the problem that the ReLU function causes neurons to not learn in the negative interval, thereby improving the recognition accuracy of the character sequence. Experimental results show that the proposed YOLOv8 and improved CRNN based ship license plate detection and recognition model has achieved excellent performance on the Guangxi Xijiang ship license plate dataset. This method can not only quickly and accurately locate the ship license plate position, but also read the digital and letter information on the ship license plate with high precision, providing strong technical support for inland waterway shipping management.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Ship License Plate Detection and Recognition Algorithm Based on YOLOv8 and Improved CRNN


    Beteiligte:
    Lin, Hongquan (Autor:in) / Liao, Weize (Autor:in) / Mo, Jinming (Autor:in) / Ye, Changyu (Autor:in) / Ji, Xiaoyu (Autor:in)


    Erscheinungsdatum :

    20.12.2024


    Format / Umfang :

    651907 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Ship Detection Based on Improved YOLOv8 Algorithm

    Cao, Xintong / Shen, Jiayu / Wang, Tao et al. | IEEE | 2024


    Real-Time Multi-class Helmet Violation Detection Using YOLOv8 with License Plate Recognition

    Charef, Ayoub / Jarir, Zahi / Quafafou, Mohamed | Springer Verlag | 2024


    Maritime Ship Tracking Algorithm Based on Improved YOLOv8 and StrongSORT

    Wu, Chen / Song, Yucun / Ge, Quanbo | IEEE | 2024



    Vietnam Vehicle Number Recognition Based on an Improved CRNN with Attention Mechanism

    Dang, Le Tran Anh / Ngoc, Vong Duong / Thien Vu, Pham Cung Le et al. | Springer Verlag | 2024