The invention discloses a partially synchronous federated learning vehicle re-identification method and system, and the method comprises the steps: enabling a plurality of local ends to obtain the respective vehicle image data, downloading the global parameters of an s-th global model from a cloud end to the local, and carrying out the local training through the combination of a local model and a local classifier; each local end performs homomorphic encryption on the local parameters after local training and uploads the local parameters to the cloud end; the cloud decrypts the local parameters, performs weighted averaging on the local parameters, generates global parameters of the (s + 1) th round, and continues to perform federated learning of the next round until the convergence requirement is met; and each local end identifies the to-be-identified vehicle image data by using the trained local model to obtain the multi-angle appearance characteristics of the vehicle, and matches the multi-angle appearance characteristics with the registered vehicle information to determine the owner information of the illegal parking vehicle. According to the invention, an algorithm platform for protecting data privacy and saving the computing power of a property management company is established.

    本发明公开了一种部分同步的联邦学习车辆重识别方法及系统,方法包括:多个本地端获取各自的车辆图像数据,并从云端下载第s轮的全局模型的全局参数至本地,并结合本地模型与本地的分类器进行本地训练;各所述本地端将本地训练后的本地参数进行同态加密,并上传至云端;云端对各所述本地参数进行解密,并对各所述本地参数进行加权平均,生成第s+1轮的全局参数,并继续进行下一轮的联邦学习,直至达到收敛要求;各本地端利用训练后的本地模型对待识别的车辆图像数据进行识别,得到车辆的多角度外观特征,并将所述多角度外观特征与登记的车辆信息进行匹配,确定违停的车辆车主信息。本发明搭建了一个保护数据隐私、节约物业公司算力的算法平台。


    Access

    Download


    Export, share and cite



    Title :

    Partially synchronous federated learning vehicle re-identification method and system


    Additional title:

    一种部分同步的联邦学习车辆重识别方法及系统


    Contributors:
    CHANG KEXIN (author) / YUAN JI (author)

    Publication date :

    2022-11-11


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06V / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / H04L TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION , Übertragung digitaler Information, z.B. Telegrafieverkehr



    Fast Wireless Federated Learning with Adaptive Synchronous Degree Control

    Chen, Zhixiong / Yi, Wenqiang / Shin, Hyundong et al. | IEEE | 2024


    Electric vehicle charging facility recommendation method and system based on federated learning

    WANG XIAOHUI / AN NINGYU / ZHENG XIAOKUN et al. | European Patent Office | 2020

    Free access


    Distributed Unknown Specific Emitter Identification Based on Federated Learning

    Xiao, Hongyujie / Liu, Heng / Zhou, Yi et al. | IEEE | 2024


    Vehicle trajectory prediction method based on air calculation and federated learning

    WANG XINYI / FAN YUXIN / XIA FANGHAO et al. | European Patent Office | 2024

    Free access