本发明公开了一种基于深度学习的机动车斑马线礼让行人自动审核方法:获取卡口摄像头抓拍到的车辆经过交通路口时的照片;使用行人检测算法检测图片上是否有行人;使用行人重识别算法追踪三张图片上的行人,并根据行人的位置判断行人是否有明显位移;使用场景分割算法分割出人行横道区域的信息,并判断行人是否在人行横道区域;使用车辆检测算法检测出抓拍图上的车辆;使用车牌识别和车辆重识别算法定位并追踪三种图上待审核的车辆;判断追踪的车辆行驶方向与行人位移方向是否有冲突;判断第一张照片是否越过停止线,且第二、三张图片机动车是否有位移并阻挡了行人通过。

    The invention discloses a motor vehicle zebra crossing pedestrian-yielding automatic auditing method based on deep learning. The motor vehicle zebra crossing pedestrian-yielding automatic auditing method comprises the steps of: acquiring pictures of a vehicle passing by means of a traffic intersection captured by a means of a checkpoint camera; using a pedestrian detection algorithm to detect whether a pedestrian exists in the picture; tracking pedestrians on three pictures by using a pedestrian re-identification algorithm, and judging whether the pedestrians have obvious displacement or not according to positions of the pedestrians; using a scene segmentation algorithm to segment information of a pedestrian crossing region, and judging whether the pedestrians are in the pedestrian crossing region or not; detecting the vehicle on the snapshot pictures by using a vehicle detection algorithm; positioning and tracking a to-be-audited vehicle in the three pictures by using a license platerecognition algorithms and a vehicle re-recognition algorithm; judging whether a driving direction of the tracked vehicle conflicts with a pedestrian displacement direction or not; and judging whetherthe vehicle crosses a stop line or not in the first picture and whether the motor vehicle has displacement and stops the pedestrians from passing in the second picture and the third picture.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    一种基于深度学习的机动车斑马线礼让行人自动审核方法


    Erscheinungsdatum :

    2022-08-30


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06V