With the promulgation of the "Made in China 2025" program, State Grid Corporation of China proposes to build a modern smart supply chain system with digital, intelligent, networked, and standardized characteristics, to achieve precise control of physical resources and ensure the consistency of "accounts, cards, and materials". This article proposes electricity distribution network material images recognition method based on MSA-YOLOv5x to address the issues of low detection accuracy and slow speed in existing distribution network material recognition methods. The ACON-C activation function is introduced to improve the feature extraction ability of the network. Y-SPP module is introduced to combine different channel information, effectively integrate different receptive field features, and enhance the feature representation capability of the network. The MSA module is introduced to enhance the ability of capturing the global feature relationship and improve the detection accuracy of the algorithm. Focal-EIoU was used to accelerate the convergence of the model and solve the problem of sample imbalance in bounding box regression. The experimental results show that when the threshold is 0.5, the mAP of the MSA-YOLOv5x reaches 97.7%, which is 6.4% higher than that of the original YOLOv5x model. The detection speed of the improved model is 62 frames/s, which is faster than that of the original YOLOv5x model. The method proposed in this article can effectively detect materials in electricity distribution networks in complex work scenarios, which can help State Grid operators accurately control materials and reduce the intensity of human monitoring work.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Research on the Recognition Algorithm of Typical Electricity Distribution Network Material Images Based on MSA-YOLOv5x


    Contributors:
    Mo, Jiajie (author) / Wang, Yong (author) / Wang, Liujun (author) / Chen, Yu (author) / Gao, Zhan (author)


    Publication date :

    2023-10-11


    Size :

    4194886 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Wood Defect Inspection on Dead Knots and Pinholes Using YOLOv5x Algorithm

    Yi, Liew Pei / Akbar, Muhammad Firdaus / Rosdi, Bakhtiar Affendi et al. | Springer Verlag | 2024




    RAILWAY VEHICLE HAVING AN OPTIMIZED ELECTRICITY DISTRIBUTION NETWORK

    DELEMOTTE JEAN-MICHEL | European Patent Office | 2018

    Free access