WO2022099608A1 - Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route - Google Patents

Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route Download PDF

Info

Publication number
WO2022099608A1
WO2022099608A1 PCT/CN2020/128660 CN2020128660W WO2022099608A1 WO 2022099608 A1 WO2022099608 A1 WO 2022099608A1 CN 2020128660 W CN2020128660 W CN 2020128660W WO 2022099608 A1 WO2022099608 A1 WO 2022099608A1
Authority
WO
WIPO (PCT)
Prior art keywords
parameter
feature
vector
input
dimensional
Prior art date
Application number
PCT/CN2020/128660
Other languages
English (en)
Chinese (zh)
Other versions
WO2022099608A9 (fr
Inventor
金序能
Original Assignee
金序能
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 金序能 filed Critical 金序能
Priority to PCT/CN2020/128660 priority Critical patent/WO2022099608A1/fr
Publication of WO2022099608A1 publication Critical patent/WO2022099608A1/fr
Publication of WO2022099608A9 publication Critical patent/WO2022099608A9/fr

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled

Definitions

  • the present application relates to the technical field of data processing, and in particular, to a method for obtaining attribute categories of traffic accidents on expressways.
  • the existing navigation software can only prompt the accident ahead and the approximate distance of the traffic jam, but cannot predict the duration of the accident.
  • the embodiments of the present application provide a method for obtaining attribute categories of traffic accidents on expressways, so as to at least solve the problem that the existing navigation software in the related art cannot predict the accident duration, and there is no effective solution at present.
  • a method for acquiring attribute categories of traffic accidents on expressways including: acquiring a weight map corresponding to different parameter features of a target object in an input image, wherein the weight map is used to represent different The weight corresponding to the parameter feature, the target object includes the party involved in the traffic accident, the vehicle involved, and related items; the input vector of the parameter feature is obtained according to the weight map of the parameter feature, wherein the parameter feature includes: The number of parties, the disability level of the parties, the number of vehicles involved, the damage level of the vehicles involved, the value of the related items, and the degree of damage to the related items; input the input vector of the parameter feature into the attribute classifier to obtain the attribute category corresponding to the parameter feature , wherein the attribute category is used to indicate the severity level of the traffic accident contained in the input image.
  • the extracting weight maps corresponding to different parameter features of the target object in the input image includes: extracting general features of the target object to obtain a global three-dimensional feature vector, wherein the general features include one or more of the parametric features; according to the global three-dimensional feature vector, extract the parametric three-dimensional feature vector corresponding to the parametric feature, and convert the parametric three-dimensional feature vector into a parametric one-dimensional feature vector; according to the The parameter one-dimensional feature vector and the target three-dimensional feature vector generate a weight map corresponding to the parameter feature, wherein the target three-dimensional feature vector is the three-dimensional feature vector corresponding to the parameter feature with the highest complexity among the parameter features of the target object.
  • the method before extracting the general feature of the target object to obtain a global three-dimensional feature vector, the method further includes: acquiring a plurality of the input images shot by cameras with different angles at the same moment; identifying each of the input images In the area where the target object is located, a plurality of the input images at the same time are integrated into a panoramic image, and the repeated target objects are fused; the size of the input image is adjusted, wherein the adjusted The input image contains all the target objects.
  • extracting a parameter three-dimensional feature vector corresponding to the parameter feature according to the global three-dimensional feature vector, and converting the parameter three-dimensional feature vector into a parameter one-dimensional feature vector includes: converting the global three-dimensional feature vector.
  • the three-dimensional feature vector is input into a 2*2 convolutional layer, and the parameter three-dimensional feature vector corresponding to the parameter feature is obtained; the parameter three-dimensional feature vector is input into the fully connected layer, and the loss function is used to train the parameter three-dimensional feature vector, A one-dimensional feature vector of the parameters is obtained.
  • generating the weight map corresponding to the parameter feature according to the parameter one-dimensional feature vector and the target three-dimensional feature vector includes: passing the one-dimensional feature vector and the target three-dimensional feature vector extracted in the parameter feature extraction module through Attribute-related attention guidance to get a weight map.
  • the obtaining the input vector of the parameter feature according to the weight map of the parameter feature includes: obtaining the input vector of the parameter feature according to the weight map of the parameter feature and the target three-dimensional feature vector.
  • the obtaining the input vector of the parameter feature according to the weight map of the parameter feature includes: when the first parameter feature is correlated with the second parameter feature and the third parameter feature and the first parameter feature is complex.
  • the input vector of the first parameter feature is obtained by the following formula:
  • f a2 f a1 + ⁇ f b2 + ⁇ f c2 (f a1 ,f b2 ,f c2 ⁇ R d )
  • f a2 represents the input vector of the first parameter feature
  • f a1 represents the parameter one-dimensional feature vector of the first parameter feature
  • f b2 represents the input vector of the second parameter feature
  • f c2 represents the The input vector of the third parameter feature
  • R d represents a 1 ⁇ d-dimensional vector in the real number domain
  • ⁇ ( ⁇ (0,1)) is a constant, indicating the relationship between the second parameter feature and the first parameter feature
  • Correlation coefficient, ⁇ ( ⁇ (0,1)) is a constant, the correlation coefficient between the third parameter feature and the first parameter feature.
  • the method further includes: according to the attribute category corresponding to the parameter feature and the weight of different parameter features , and determine the processing time required for the traffic accident contained in the input image.
  • a computer-readable storage medium is also provided, where a computer program is stored in the storage medium, wherein the computer program is configured to execute any one of the above method embodiments when running steps in .
  • an electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute Steps in any one of the above method embodiments.
  • the weight map corresponding to different parameter features of the target object in the input image is obtained, and the target object includes the party involved in the traffic accident, the vehicle involved, and related items, and the input vector of the parameter feature is obtained according to the weight map of the parameter feature, and the step
  • the input vector of the parameter feature is input into the attribute classifier, and the attribute category corresponding to the parameter feature is obtained, wherein the attribute category is used to indicate the severity level of the traffic accident contained in the input image, which solves the problem that the existing navigation software in the prior art is unable to detect the traffic accident.
  • the problem of accident duration prediction can be based on the severity of the accident and the processing time of similar accidents in the past to predict the estimated duration of the current traffic accident.
  • FIG. 1 is a flowchart of an optional method for acquiring attribute categories of traffic accidents on expressways according to an embodiment of the present application.
  • FIG. 1 is a flowchart of an optional method for acquiring attribute categories of traffic accidents on expressways according to an embodiment of the present application. As shown in FIG. 1 , the method includes:
  • Step S102 obtaining weight maps corresponding to different parameter features of the target object in the input image, wherein the weight map is used to represent the weights corresponding to different parameter features, and the target objects include the parties involved in the traffic accident, the vehicles involved, and related items;
  • Step S104 obtaining the input vector of the parameter feature according to the weight map of the parameter feature, wherein the parameter feature includes: the number of parties involved, the disability level of the party involved, the number of vehicles involved, the damage level of the vehicle involved, the value of related items, and the degree of damage to the related items;
  • Step S106 Input the input vector of the parameter feature into the attribute classifier to obtain the attribute category corresponding to the parameter feature, wherein the attribute category is used to indicate the severity level of the traffic accident contained in the input image.
  • the extracting weight maps corresponding to different parameter features of the target object in the input image includes: extracting general features of the target object to obtain a global three-dimensional feature vector, wherein the general features include one or more of the parametric features; according to the global three-dimensional feature vector, extract the parametric three-dimensional feature vector corresponding to the parametric feature, and convert the parametric three-dimensional feature vector into a parametric one-dimensional feature vector; according to the The parameter one-dimensional feature vector and the target three-dimensional feature vector generate a weight map corresponding to the parameter feature, wherein the target three-dimensional feature vector is the three-dimensional feature vector corresponding to the parameter feature with the highest complexity among the parameter features of the target object.
  • the method before extracting the general feature of the target object to obtain a global three-dimensional feature vector, the method further includes: acquiring a plurality of the input images shot by cameras with different angles at the same moment; identifying each of the input images In the area where the target object is located, a plurality of the input images at the same time are integrated into a panoramic image, and the repeated target objects are fused; the size of the input image is adjusted, wherein the adjusted The input image contains all the target objects.
  • extracting a parameter three-dimensional feature vector corresponding to the parameter feature according to the global three-dimensional feature vector, and converting the parameter three-dimensional feature vector into a parameter one-dimensional feature vector includes: converting the global three-dimensional feature vector.
  • the three-dimensional feature vector is input into a 2*2 convolutional layer, and the parameter three-dimensional feature vector corresponding to the parameter feature is obtained; the parameter three-dimensional feature vector is input into the fully connected layer, and the loss function is used to train the parameter three-dimensional feature vector, A one-dimensional feature vector of the parameters is obtained.
  • generating the weight map corresponding to the parameter feature according to the parameter one-dimensional feature vector and the target three-dimensional feature vector includes: passing the one-dimensional feature vector and the target three-dimensional feature vector extracted in the parameter feature extraction module through Attribute-related attention guidance to get a weight map.
  • the obtaining the input vector of the parameter feature according to the weight map of the parameter feature includes: obtaining the input vector of the parameter feature according to the weight map of the parameter feature and the target three-dimensional feature vector.
  • the obtaining the input vector of the parameter feature according to the weight map of the parameter feature includes: when the first parameter feature is correlated with the second parameter feature and the third parameter feature and the first parameter feature is complex.
  • the input vector of the first parameter feature is obtained by the following formula:
  • f a2 f a1 + ⁇ f b2 + ⁇ f c2 (f a1 ,f b2 ,f c2 ⁇ R d )
  • f a2 represents the input vector of the first parameter feature
  • f a1 represents the parameter one-dimensional feature vector of the first parameter feature
  • f b2 represents the input vector of the second parameter feature
  • f c2 represents the The input vector of the third parameter feature
  • R d represents a 1 ⁇ d-dimensional vector in the real number domain
  • ⁇ ( ⁇ (0,1)) is a constant, indicating the relationship between the second parameter feature and the first parameter feature
  • Correlation coefficient, ⁇ ( ⁇ (0,1)) is a constant, the correlation coefficient between the third parameter feature and the first parameter feature.
  • the method further includes: according to the attribute category corresponding to the parameter feature and the weight of different parameter features , and determine the processing time required for the traffic accident contained in the input image.
  • the number of parties involved in the traffic accident, the vehicles involved and related items, the number of parties involved, the level of disability of the parties, the number of vehicles involved, and the level of damage to the vehicles involved can be determined , the value of related items, and the degree of damage to related items to determine the category of the current traffic accident and the corresponding severity level, and combine the category and processing time of the previous traffic accident stored in the database to predict the duration of the current traffic accident, and then Get an estimate of the duration of the traffic jam.
  • the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is better implementation.
  • the technical solution of the present application can be embodied in the form of a software product in essence or in a part that contributes to the prior art, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, CD-ROM), including several instructions to make a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) execute the methods described in the various embodiments of this application.
  • a storage medium such as ROM/RAM, magnetic disk, CD-ROM
  • the storage medium may include: a flash disk, a read-only memory (Read-Only Memory, ROM), a random access device (Random Access Memory, RAM), a magnetic disk or an optical disk, and the like.
  • the integrated units in the above-mentioned embodiments are implemented in the form of software functional units and sold or used as independent products, they may be stored in the above-mentioned computer-readable storage medium.
  • the technical solution of the present application can be embodied in the form of a software product in essence, or the part that contributes to the prior art, or all or part of the technical solution, and the computer software product is stored in a storage medium,
  • Several instructions are included to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.
  • the disclosed clients may be implemented in other manners.
  • the apparatus embodiments described above are only illustrative, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components may be combined or Integration into another system, or some features can be ignored, or not implemented.
  • the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of units or modules, and may be in electrical or other forms.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Traffic Control Systems (AREA)

Abstract

Un mode de réalisation de la présente demande concerne un procédé d'acquisition d'une catégorie d'attribut d'un accident de la circulation sur une route. Le procédé consiste : à acquérir un graphe pondéré correspondant à différentes caractéristiques de paramètre d'objets cibles dans une image d'entrée, les objets cibles comprenant des parties et des véhicules impliqués dans un accident de la circulation, ainsi que des articles connexes ; à acquérir, en fonction du graphe pondéré des caractéristiques de paramètre, des vecteurs d'entrée des caractéristiques de paramètre ; et à entrer les vecteurs d'entrée des caractéristiques de paramètre dans un classificateur d'attributs, de manière à obtenir une catégorie d'attribut correspondant aux caractéristiques de paramètre, la catégorie d'attribut servant à indiquer un niveau de gravité de l'accident de la circulation contenu dans l'image d'entrée. L'invention résout le problème de l'état de la technique d'impossibilité de prédiction de la durée d'un accident par un logiciel de navigation existant.
PCT/CN2020/128660 2020-11-13 2020-11-13 Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route WO2022099608A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/128660 WO2022099608A1 (fr) 2020-11-13 2020-11-13 Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/128660 WO2022099608A1 (fr) 2020-11-13 2020-11-13 Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route

Publications (2)

Publication Number Publication Date
WO2022099608A1 true WO2022099608A1 (fr) 2022-05-19
WO2022099608A9 WO2022099608A9 (fr) 2022-07-14

Family

ID=81602025

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/128660 WO2022099608A1 (fr) 2020-11-13 2020-11-13 Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route

Country Status (1)

Country Link
WO (1) WO2022099608A1 (fr)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN202075862U (zh) * 2011-03-28 2011-12-14 长安大学 一种高速公路交通事故影响预测与控制系统
CN103258432A (zh) * 2013-04-19 2013-08-21 西安交通大学 基于视频的交通事故自动识别处理方法和系统
US20160272140A1 (en) * 2015-03-16 2016-09-22 Hyundai Motor Company Device and method for providing vehicle accident information
CN109034264A (zh) * 2018-08-15 2018-12-18 云南大学 交通事故严重性预测csp-cnn模型及其建模方法
CN110276959A (zh) * 2019-06-26 2019-09-24 奇瑞汽车股份有限公司 交通事故的处理方法、装置及存储介质
CN111859291A (zh) * 2020-06-23 2020-10-30 北京百度网讯科技有限公司 交通事故识别方法、装置、设备和计算机存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN202075862U (zh) * 2011-03-28 2011-12-14 长安大学 一种高速公路交通事故影响预测与控制系统
CN103258432A (zh) * 2013-04-19 2013-08-21 西安交通大学 基于视频的交通事故自动识别处理方法和系统
US20160272140A1 (en) * 2015-03-16 2016-09-22 Hyundai Motor Company Device and method for providing vehicle accident information
CN109034264A (zh) * 2018-08-15 2018-12-18 云南大学 交通事故严重性预测csp-cnn模型及其建模方法
CN110276959A (zh) * 2019-06-26 2019-09-24 奇瑞汽车股份有限公司 交通事故的处理方法、装置及存储介质
CN111859291A (zh) * 2020-06-23 2020-10-30 北京百度网讯科技有限公司 交通事故识别方法、装置、设备和计算机存储介质

Also Published As

Publication number Publication date
WO2022099608A9 (fr) 2022-07-14

Similar Documents

Publication Publication Date Title
US10789786B2 (en) Picture-based vehicle loss assessment
JP6873237B2 (ja) 画像ベースの車両損害評価方法、装置、およびシステム、ならびに電子デバイス
CN109271418B (zh) 可疑团伙识别方法、装置、设备及计算机可读存储介质
CN111886842A (zh) 使用基于阈值的匹配进行远程用户身份验证
CN110490066B (zh) 基于图片分析的目标检测方法、装置及计算机设备
WO2019167784A1 (fr) Dispositif de spécification de position, procédé de spécification de position et programme informatique
CN111310770A (zh) 目标检测方法和装置
CN111242119A (zh) 车辆图像的处理方法、装置、电子设备和计算机可读介质
US20200136818A1 (en) System for generating personalized service content
CN114663871A (zh) 图像识别方法、训练方法、装置、系统及存储介质
CN110823228B (zh) 一种路径生成方法及装置
US9514368B2 (en) Contextual information of visual media
CN116152870A (zh) 人脸识别方法、装置、电子设备及计算机可读存储介质
CN115131634A (zh) 图像识别方法、装置、设备、存储介质及计算机程序产品
CN108170751A (zh) 用于处理图像的方法和装置
WO2022099608A1 (fr) Procédé d'acquisition de catégorie d'attribut d'un accident de la circulation sur une route
JP6542819B2 (ja) 画像監視システム
CN111192150B (zh) 车辆出险代理业务的处理方法、装置、设备及存储介质
CN111985531A (zh) 异常资源需求集群的确定方法、装置、设备和存储介质
WO2019187107A1 (fr) Dispositif de traitement d'informations, procédé de commande et programme
CN115731451A (zh) 模型训练方法、装置、电子设备及存储介质
CN114140744A (zh) 基于对象的数量检测方法、装置、电子设备及存储介质
CN115082873A (zh) 基于通路融合的图像识别方法、装置及存储介质
CN111209807A (zh) 一种基于yolov3的视频结构化方法及系统
CN109785157A (zh) 一种基于人脸识别的车辆定损方法、存储介质及服务器

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20961164

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20961164

Country of ref document: EP

Kind code of ref document: A1