CN113591730A - Method, device and equipment for recognizing lane grouping line - Google Patents

Method, device and equipment for recognizing lane grouping line Download PDF

Info

Publication number
CN113591730A
CN113591730A CN202110886471.7A CN202110886471A CN113591730A CN 113591730 A CN113591730 A CN 113591730A CN 202110886471 A CN202110886471 A CN 202110886471A CN 113591730 A CN113591730 A CN 113591730A
Authority
CN
China
Prior art keywords
lane
lane line
line
data set
detected
Prior art date
Legal status (The legal status 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 status listed.)
Granted
Application number
CN202110886471.7A
Other languages
Chinese (zh)
Other versions
CN113591730B (en
Inventor
潘皓博
王志勇
李兴义
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Langge Technology Co ltd
Original Assignee
Hubei Ecarx Technology Co Ltd
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 Hubei Ecarx Technology Co Ltd filed Critical Hubei Ecarx Technology Co Ltd
Priority to CN202110886471.7A priority Critical patent/CN113591730B/en
Publication of CN113591730A publication Critical patent/CN113591730A/en
Application granted granted Critical
Publication of CN113591730B publication Critical patent/CN113591730B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Traffic Control Systems (AREA)
  • Image Analysis (AREA)

Abstract

The method, the device and the equipment for recognizing the lane grouping line input the lane line characteristics of the lane line to be detected, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected into a trained machine learning model by extracting the lane line characteristics of the lane line to be detected, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected, and output a prediction result, wherein the prediction result comprises the lane grouping line. The extracted lane line characteristics, the characteristics of the road elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road elements around the lane line to be detected are used for training the machine learning model, so that when the characteristics are input into the trained machine learning model, the lane grouping line in the lane line to be detected can be obtained in time, and the lanes can be efficiently and comprehensively grouped.

Description

Method, device and equipment for recognizing lane grouping line
Technical Field
The application relates to the technical field of automobiles, in particular to a method, a device and equipment for recognizing lane grouping lines.
Background
The high-precision map is a navigation and positioning electronic map refined to the level of a lane line, and is indispensable data in the current automatic driving system.
As an important element in a high-precision map, a lane line not only needs to have high precision, but also needs to be manufactured in strict compliance with certain specifications. For example, on lanes running in the same direction, it is necessary to break and number lane lines. An important link of the uniform breaking and numbering of the lane lines is to group the lane lines, namely dividing lanes on the road into different lane line groups according to the distinguishing modes of up/down, main and auxiliary roads, physical isolation facilities and the like.
The conventional lane grouping method is through manual work or rule extraction. However, the manual operation mode has the problems of high drawing cost, low efficiency and the like, so that the map updating speed is delayed, and the timeliness of the map is reduced. The rule extraction mode also has limitation in a large-range drawing task, the larger the drawing range is, the more complex the scene is, and the required rules are difficult to maintain in quantity or logic. Thus, it is difficult for the conventional lane grouping method to efficiently and comprehensively group lanes.
Disclosure of Invention
The application provides a method, a device and equipment for recognizing lane grouping lines, which are used for solving the problem that the traditional lane grouping method is difficult to efficiently and comprehensively group lanes.
In a first aspect, the present application provides a method of identifying lane grouping lines, the method comprising:
extracting corresponding characteristics of the lane line to be detected, wherein the corresponding characteristics comprise: the lane line characteristics, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected;
and inputting the corresponding characteristics of the lane line to be detected into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises a lane grouping line.
Optionally, the lane line feature includes: at least one of the mark, the geometric length, the direction, the type and the track distance of the lane line to be detected;
the characteristics of the road surface elements around the lane line to be measured include: the type of the arrow on the two sides of the lane line to be detected, and/or at least one of the number, the type, the width and the direction of the lane line on the two sides of the lane line to be detected;
the lane line characteristics and the relative characteristics of the road surface elements around the lane line to be detected include: the system comprises an adjacency matrix used for identifying the topological connection relation of the lane lines to be detected, and/or the distance between the lane lines to be detected and the nearest road surface identification, and/or at least one of the relative position relation, the width ratio, the height difference of an external matrix, the curvature difference and the angle difference of the lane lines to be detected on two sides of the lane lines to be detected and adjacent to the lane lines to be detected.
Optionally, before inputting the corresponding features of the lane line to be detected into the machine learning model, the method further includes:
collecting a training data set, wherein each lane line data in the training data set is provided with a corresponding first label, and the first label comprises a grouping line or a non-grouping line;
extracting a respective feature from each lane line data in the training data set, the respective feature comprising: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line;
and inputting the first label and the corresponding features extracted from the training data set into a built machine learning model for training.
Optionally, before inputting the corresponding features of the lane line to be detected into the machine learning model, the method further includes:
collecting a test data set, wherein each lane line data in the test data set is provided with a corresponding second label, and the second label comprises a grouping line or a non-grouping line;
extracting a respective feature from each lane line data in the test data set, the respective feature comprising: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line;
and inputting the second label and the corresponding features extracted from the test data set into a trained machine learning model for verification.
Optionally, after the acquiring the training data set, the method further includes:
combining and representing the first label of each lane line data in a training data set by using a vector to obtain a label vector of the training data set;
acquiring a first feature vector of each corresponding feature in the training data set, and merging the first feature vectors of the corresponding features of each lane line to obtain a first feature matrix;
correspondingly, the training of the machine learning model established by inputting the first label and the corresponding features extracted from the training data set specifically includes:
and inputting the label vector of the training data set and the first characteristic matrix into an established machine learning model for training.
Optionally, the respective features comprise discrete features and continuous features;
the obtaining of the first feature vector of each corresponding feature in the training data set, and merging the first feature vectors of the corresponding features of each lane line to obtain the first feature matrix specifically includes:
adopting one-hot encoding processing on the discrete features to convert the discrete features into binary features, and combining the binary features into a binary vector to obtain a first feature vector of the discrete features;
normalizing the continuous features to enable all numerical values in feature vectors of the continuous features to be within a specific interval range so as to obtain first feature vectors of the continuous features;
and combining the first feature vector of the discrete features and the first feature vector of the continuous features to obtain a first feature matrix.
Optionally, after the collecting the test data set, the method further includes:
combining and representing second labels of each lane line data in a test data set by using vectors to obtain label vectors in the test data set;
acquiring a second eigenvector of each corresponding characteristic in the test data set, and merging the second eigenvectors of the corresponding characteristic of each lane line to obtain a second eigenvector matrix;
correspondingly, the inputting the second label and the corresponding features extracted from the test data set into the trained machine learning model for verification specifically includes:
and inputting the label vector of the test data set and the second feature matrix into a trained machine learning model for verification.
Optionally, the machine learning model comprises a support vector machine model;
inputting the label vector of the training set and the first feature matrix into a machine learning model established for training, and specifically comprising:
inputting the label vectors of the training set and the first feature matrix into a support vector machine model with initialized parameters for training;
inputting the label vector of the test data set and the second feature matrix into a trained machine learning model for verification, and specifically comprising:
inputting the label vector of the test data set and the second feature matrix into the trained support vector machine model, and outputting a prediction result of the test data set;
and comparing the prediction result of the test data set with the label vector of the test set.
Optionally, if a comparison result between the prediction result of the test data set and the tag vector of the test data set is greater than or equal to a preset value, the support vector machine model is saved;
and if the comparison result of the prediction result of the test data set and the label vector of the test data set is smaller than a preset value, adjusting the hyper-parameters of the support vector machine model initialized by the parameters until the comparison result is larger than or equal to the preset value.
Optionally, the data of the training data set is larger than the data of the validation data set.
Optionally, the lane line to be tested includes a lane line having a topological relationship with the lane line grouping line;
after acquiring the lane grouping line, the method further comprises:
inputting the lane line to be detected into a graph model, and outputting a grouping line having a topological relation with the lane grouping line;
and taking the lane grouping line and the grouping line having the topological relation with the lane grouping line as a final lane grouping line.
Optionally, before the lane line to be detected is input into the graph model, the method further includes:
and establishing graph model data by taking the end points of the known lane lines as vertexes and the known lane lines as sides, obtaining the connected sides through the degree of each vertex, and establishing the connection relation between the sides to obtain the graph model, wherein the known lane lines are the lane lines which are in topological connection with the lane grouping lines.
In a second aspect, the present application provides an apparatus for recognizing a lane grouping line, the apparatus comprising:
the extraction module is used for extracting corresponding characteristics of the lane line to be detected, and the corresponding characteristics comprise: the lane line characteristics, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected;
and the control module is used for inputting the corresponding characteristics of the lane line to be tested into the trained machine learning model and outputting a prediction result, wherein the prediction result comprises a lane grouping line.
In a third aspect, the present application provides an electronic device, comprising: a memory and a processor;
the memory is used for storing program instructions; the processor is configured to invoke program instructions in the memory to perform the method of identifying lane grouping lines in the first aspect and any one of the possible designs of the first aspect.
In a fourth aspect, the present application provides a readable storage medium having stored therein computer instructions, which when executed by at least one processor of an electronic device, cause the electronic device to perform the method of identifying lane grouping lines in any one of the possible designs of the first aspect and the first aspect.
In a fifth aspect, the present application provides a computer program product comprising computer instructions which, when executed by at least one processor of an electronic device, cause the electronic device to perform the method of identifying lane grouping lines in any one of the possible designs of the first aspect and the first aspect.
The method for recognizing the lane grouping line comprises the steps of extracting lane line characteristics of a lane line to be detected, characteristics of road surface elements around the lane line to be detected and relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected, inputting the lane line characteristics of the lane line to be detected, the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected into a trained machine learning model, outputting a prediction result, and outputting the prediction result, wherein the prediction result comprises the lane grouping line. The extracted lane line characteristics, the characteristics of the road elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road elements around the lane line to be detected are used for training the machine learning model, so that when the characteristics are input into the trained machine learning model, the lane grouping line in the lane line to be detected can be obtained in time, and the lanes can be efficiently and comprehensively grouped.
Drawings
In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a method for identifying lane grouping lines according to an embodiment of the present application;
FIG. 2 is a flow chart of a method for identifying lane grouping lines provided in another embodiment of the present application;
FIG. 3 is a flow chart of a method for identifying lane grouping lines provided in another embodiment of the present application;
FIG. 4 is a flow chart of a method for identifying lane grouping lines provided in another embodiment of the present application;
fig. 5 is a schematic structural diagram of an apparatus for identifying lane grouping lines according to an embodiment of the present application;
fig. 6 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present application.
Detailed Description
To make the purpose, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below with reference to the drawings in the present application, and it is obvious that the described embodiments are some, but not all embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
At present, automatic driving becomes an important industrial trend in the automobile field, but the development is slow due to more involved and covered industrial chain links, and a high-precision map becomes a key link of automatic driving development. The high-precision map is an electronic map for navigation and positioning which is refined to the level of a lane line. As an important element in a high-precision map, a lane line not only needs to have high precision, but also needs to be manufactured in strict compliance with certain specifications. For example, on lanes running in the same direction, it is necessary to break and number lane lines. The regular breaking and numbering refers to separating different roads, and then numbering the lane lines on different roads respectively, for example, the road 1 includes lane line 1 and lane line 2, and the road 2 includes lane line 1, lane line 2 and lane line 3.
The conventional lane grouping method is through manual work or rule extraction. However, the manual operation mode has the problems of high drawing cost, low efficiency and the like, so that the map updating speed is delayed, and the timeliness of the map is reduced. The rule extraction mode also has limitation in a large-range drawing task, the larger the drawing range is, the more complex the scene is, and the required rules are difficult to maintain in quantity or logic. Thus, it is difficult for the conventional lane grouping method to efficiently and comprehensively group lanes.
In view of the above problems, the present application provides a method for recognizing lane grouping lines, in which corresponding features of a lane line to be detected are extracted, and the corresponding features include: inputting the corresponding characteristics of the lane line to be detected into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises a lane grouping line. Because the machine learning model is trained through the corresponding characteristics of the lane grouping lines and the corresponding characteristics of the lane non-grouping lines, when the corresponding characteristics of the lane line to be detected are input into the trained machine learning model, the lane grouping lines in the lane line to be detected can be obtained in time, and therefore the lanes can be efficiently and comprehensively grouped.
The technical solution of the present application will be described in detail below with specific examples. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
The server extracts lane line characteristics of a lane line to be detected, characteristics of road surface elements around the lane line to be detected and relative characteristics of the lane line characteristics and the surrounding road surface elements, then the server inputs the lane line characteristics of the lane line to be detected, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected into a trained machine learning model, the machine learning model outputs a prediction result, the prediction result comprises lane grouping lines, and therefore the lane grouping lines in the lane line to be detected are obtained.
Fig. 1 is a flowchart illustrating a method for identifying lane grouping lines according to an embodiment of the present application. The method of the embodiment may include the steps of:
s101, extracting corresponding characteristics of the lane line to be detected, wherein the corresponding characteristics comprise: lane line characteristics, characteristics of road surface elements around the lane line to be detected, and relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected.
The number of lane lines to be tested may be one or more, for example, the lane line to be tested may include one lane grouping line, one lane non-grouping line, one lane grouping line and one lane non-grouping line, one lane non-grouping line and a plurality of lane grouping lines, one lane grouping line and a plurality of lane grouping lines, or a plurality of lane grouping lines and a plurality of lane non-grouping lines.
The extracted lane line features of the lane line to be detected comprise lane line features of the lane group line and/or lane line features of the lane non-group line, the extracted features of the road elements around the lane line to be detected comprise features of the road elements around the lane group line and/or road elements around the lane non-group line, and the extracted relative features of the lane line to be detected and the road elements around the lane line to be detected comprise relative features of the road elements around the lane group line and/or relative features of the road elements around the lane non-group line and the non-group line.
The lane line characteristics of the lane line to be tested may include at least one of an identification, a geometric length, a direction, a type, and a trajectory distance of the lane line to be tested. Including, for example, identification and geometric length. The identification may be an ID, for example. The geometric length may be the path length of the lane line to be measured. The direction may be, for example, an east-west direction, a south-north direction, a south-east-north-west direction, or a north-east-south-west direction. The type may be, for example, a solid white line, a solid single yellow line, a solid double yellow line, or a solid dashed yellow line. The track distance refers to the distance between two end points of the lane line to be detected.
The pavement elements around the lane line to be tested include: the type of the arrow on both sides of the lane line to be tested, and/or at least one of the number, type, width, and direction of the lane line on both sides of the lane line to be tested. For example, the number of lane lines on both sides of the lane line to be tested may be included, the types of arrows on both sides of the lane line to be tested may be included, and the number of lane lines on both sides of the lane line to be tested and the types of arrows on both sides of the lane line to be tested may be included. The type of arrow may be, for example, straight, left-handed, right-handed, u-turned, straight + left-handed, straight + right-handed, etc.
The lane line characteristics and the relative characteristics of the road surface elements around the lane line to be measured include: the system comprises an adjacency matrix used for representing the topological connection relation of the lane line to be detected, and/or the distance between the lane line to be detected and the nearest road surface mark, and/or at least one of the relative position relation, the width ratio, the height difference of the external matrix, the curvature difference and the angle difference of the lane lines at the two sides of the lane line to be detected and adjacent to the lane line to be detected. For example, the system may include an adjacency matrix for representing the topological connection relationship of the lane line to be tested, and may also include the distance between the lane line to be tested and the nearest road surface identifier, and the relative position relationship and the width ratio of the lane lines on both sides of the lane line to be tested and adjacent to the lane line to be tested.
The adjacent matrix used for identifying the topological connection relation of the lane lines to be detected is used for establishing the connection relation between edges by taking the end points of the lane lines to be detected as vertexes and the lane lines to be detected as edges and obtaining the connected edges through the degree of each vertex. The marking of the road surface nearest to the lane line to be tested may be, for example, a curb stone, a building, or a sign nearest to the lane line to be tested. The lane lines on both sides of the lane line to be detected and adjacent to the lane line to be detected may be, for example, the lane line closest to the left side of the lane line to be detected and the lane line closest to the right side of the lane line to be detected, for convenience of description, the lane line closest to the left side of the lane line to be detected is referred to as a first lane line, and the lane line closest to the right side of the lane line to be detected is referred to as a second lane line, and accordingly, the relative position relationship, the width ratio, the external matrix height difference, the curvature difference, and the angle difference of the lane lines adjacent to the lane line to be detected on both sides of the lane line to be detected are the relative position relationship, the width ratio, the external matrix height difference, the curvature difference, and the angle difference of the first lane line and the second lane line.
S102, inputting the corresponding characteristics of the lane line to be tested into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises lane grouping lines.
Since the machine learning model utilizes the corresponding features of the lane grouping lines, as well as the corresponding features of the lane non-grouping lines, in the training process. Therefore, when the lane line characteristics of the lane line to be measured, the characteristics of the road surface elements around the lane line to be measured, and the relative characteristics of the lane line to be measured and the road surface elements around the lane line to be measured are input into the trained machine learning model, the machine learning model can output the lane grouping line and the lane non-grouping line in the lane line to be measured.
According to the lane grouping line recognition method, when the lane line characteristics of the lane line to be detected, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected are input into the trained machine learning model, the lane grouping line in the lane line to be detected can be obtained in time, and therefore lanes can be efficiently and comprehensively grouped.
Fig. 2 is a flowchart illustrating a method for identifying lane grouping lines according to an embodiment of the present application. The method of the embodiment may include the steps of:
s201, a training data set is collected, wherein each lane line data in the training data set is provided with a corresponding first label, and the first label comprises a grouping line or a non-grouping line.
The method comprises the steps of collecting a training data set, wherein the training data set comprises a plurality of pieces of lane line data, each piece of lane line data is provided with a corresponding first label, the first label comprises a grouping line or a non-grouping line, namely the training data set comprises lane grouping line data and lane non-grouping line data, the lane grouping line data are provided with corresponding grouping line labels, and the lane non-grouping line data are provided with corresponding grouping line labels.
S202, extracting corresponding features from each lane line data in the training data set.
And extracting corresponding features from each lane line data in the training data set, namely extracting corresponding features of a grouping line from each lane grouping line, and extracting corresponding features of a non-grouping line from each non-grouping line. The corresponding features here include: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line. The current lane line is understood herein to be any one lane line in the training set.
The lane line features herein include: at least one of an identification, a geometric length, a direction, a type, and a trajectory distance of the current lane line; the characteristics of the road surface elements around the current lane line include: the type of the arrow on both sides of the current lane line, and/or at least one of the number, type, width, and direction of the lane lines on both sides of the current lane line; the relative characteristics of the lane line characteristics and the road surface elements around the current lane line include: the system comprises an adjacency matrix used for identifying the topological connection relation of the current lane line, and/or the distance between the current lane line and the nearest road surface identification, and/or at least one of the relative position relation, the width ratio, the height difference of the circumscribed matrix, the curvature difference and the angle difference of the lane lines on two sides of the current lane line and adjacent to the current lane line.
S203, inputting the first label and the corresponding features extracted from the training data set into a machine learning model established for training.
The lane line characteristics of the grouped lines and the lane line characteristics of the non-grouped lines are input into the established machine learning model for training, and the first labels are input into the machine learning model at the same time.
S204, extracting corresponding characteristics of the lane line to be detected, wherein the corresponding characteristics comprise: lane line characteristics, characteristics of road surface elements around the lane line to be detected, and relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected.
S205, inputting the corresponding characteristics of the lane line to be tested into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises lane grouping lines.
In this embodiment, steps S201 and S204 are not limited by the described operation sequence, and steps S201 and S204 may be performed in other sequences or simultaneously.
Step S204 is similar to the implementation manner of S101, and step S205 is similar to the implementation manner of S102, which is not described herein again.
According to the lane grouping line recognition method, the corresponding characteristics of the lane grouping lines and the corresponding grouping line labels, and the corresponding characteristics of the lane non-grouping lines and the corresponding non-grouping line labels are input into the machine learning model for training, so that the machine learning model can output the lane grouping lines in the lane lines to be detected according to the lane line characteristics of the lane lines to be detected, the characteristics of the road elements around the lane lines to be detected, and the relative characteristics of the lane lines to be detected and the road elements around the lane lines to be detected, and the lanes can be efficiently and comprehensively grouped.
Fig. 3 is a flowchart illustrating a method for identifying lane grouping lines according to an embodiment of the present application. The method of the embodiment may include the steps of:
s301, a training data set is collected, each lane line data in the training data set is provided with a corresponding first label, and the first label comprises a grouping line or a non-grouping line.
S302, extracting corresponding characteristics from each lane line data in the training data set.
S303, inputting the first label and the corresponding features extracted from the training data set into a machine learning model established for training.
Step S301 is similar to step S201, step S302 is similar to step S202, and step S303 is similar to step S203, which is not described herein again.
S304, collecting a test data set, wherein each lane line data in the test data set is provided with a corresponding second label, and the second label comprises a grouping line or a non-grouping line.
And collecting a test data set, wherein the test data set comprises a plurality of pieces of lane line data, each piece of lane line data is provided with a corresponding second label, and the second label comprises a grouping line or a non-grouping line, namely the test data set comprises lane grouping line data and lane non-grouping line data.
S305, extracting corresponding features from each lane line data in the test data set, wherein the corresponding features comprise: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line.
Extracting a corresponding feature from each lane line data in the test data set, where the corresponding feature includes: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line. The current lane line is understood to be any lane line in the test data set.
The lane line features herein include: at least one of an identification, a geometric length, a direction, a type, and a trajectory distance of the current lane line; the characteristics of the road surface elements around the current lane line include: the type of the arrow on both sides of the current lane line, and/or at least one of the number, type, width, and direction of the lane lines on both sides of the current lane line; the relative characteristics of the lane line characteristics and the road surface elements around the current lane line include: the system comprises an adjacency matrix used for identifying the topological connection relation of the current lane line, and/or the distance between the current lane line and the nearest road surface identification, and/or at least one of the relative position relation, the width ratio, the height difference of the circumscribed matrix, the curvature difference and the angle difference of the lane lines on two sides of the current lane line and adjacent to the current lane line.
S306, inputting the second label and the corresponding features extracted from the test data set into the trained machine learning model for verification.
And inputting the corresponding characteristics of the grouping lines, the corresponding characteristics of the non-grouping lines and the corresponding second labels in the test data set into the trained machine learning model for verification, outputting a prediction result by the machine learning model, and comparing the prediction result with the second labels to obtain the effect of the machine learning model on the test set. If the comparison result is greater than or equal to the preset precision, for example, the accuracy of the prediction result is greater than 95%, it is indicated that the machine learning model can be used for recognizing the lane grouping line, and if the comparison result is less than the preset precision, it is indicated that the machine learning model needs to be adjusted. In this embodiment, the data of the test data set may be larger than the data of the test data set, for example, three or four times as much data as the training data set.
S307, extracting corresponding features of the lane line to be detected, wherein the corresponding features comprise: lane line characteristics, characteristics of road surface elements around the lane line to be detected, and relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected.
And S308, inputting the corresponding characteristics of the lane line to be detected into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises lane grouping lines.
Step S307 is similar to the implementation manner of S101, and step S308 is similar to the implementation manner of S102, which is not described herein again.
In this embodiment, steps S301 and S307 are not limited by the described operation sequence, and steps S301 and S307 may be performed in other sequences or simultaneously.
According to the lane grouping line identification method, after the machine learning model is trained by using the corresponding features in the training data set, the machine learning model is verified by using the corresponding features in the verification data set, so that the prediction result of the machine learning model is improved.
Fig. 4 is a flowchart illustrating a method for identifying lane grouping lines according to an embodiment of the present application. The method of the embodiment may include the steps of:
s401, a training data set is collected, each lane line data in the training data set is provided with a corresponding first label, and the first label comprises a grouping line or a non-grouping line.
S402, extracting corresponding features from each lane line data in the training data set, wherein the corresponding features comprise: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line.
And S403, combining and representing the first labels of each lane line data in the training data set by using a vector to obtain the label vector of the training data set.
The first label combination of each lane line data in the training data set is represented by a vector, and the dimension of the vector is the number of lane lines in the training data set, for example, the training data set includes five lane lines, which are represented as a first lane line, a second lane line, a third lane line, a fourth lane line and a fifth lane line, the first lane line is a lane grouping line, the second lane line is a lane grouping line, the third lane line is a lane non-grouping line, the fourth lane line is a lane non-grouping line, the fifth lane line is a lane grouping line, the labels are quantified, the grouping line label is represented as 1, and the non-grouping line label is represented as 0, so that the form of the five lane lines after combination and representation by the vector is {1, 1,0, 0, 1}, i.e., the label vector of the training data set is {1, 1,0, 0, 1 }.
In this embodiment, steps S402 and S403 are not limited by the described operation sequence, and steps S402 and S403 may be performed in other sequences or simultaneously.
S404, obtaining a first feature vector of each corresponding feature in the training data set, and combining the first feature vectors of each corresponding feature to obtain a first feature matrix.
In step S402, a corresponding feature is extracted from each lane line data in the training data set, each lane line data may extract a plurality of corresponding features, and the training data set includes a plurality of lane lines, so that a plurality of corresponding features may be extracted from the training data set.
And then, acquiring a first feature vector of each corresponding feature, wherein the dimensionality of the first feature vector is the number of lane lines in the data set. For example, the respective characteristics are the number of lane lines on both sides of the current lane line in the characteristics of the road surface elements around the current lane line. Specifically, the number of lane lines on both sides of each lane line in the training data set is first calculated, for example, the number of lane lines on both sides of the first lane line is 5, the number of lane lines on both sides of the second lane line is 3, the number of lane lines on both sides of the third lane line is 4, the number of lane lines on both sides of the fourth lane line is 5, and the number of lane lines on both sides of the fifth lane line is 4. And then combining the numerical values of the number of the lane lines on the two sides of each lane line to form a vector, namely {5, 3, 4, 5, 4}, and acquiring the characteristic vector of the number of the lane lines on the two sides of the current lane line as {5, 3, 4, 5, 4 }.
Then, the first feature vectors of each corresponding feature are combined to obtain a first feature matrix. Specifically, the first eigenvectors of all corresponding characteristics are superposed in the column direction according to the characteristic combination sequence to form a first characteristic matrix, the column number of the first characteristic matrix represents the number of lane lines in the test data set, and the row number of the first characteristic matrix represents the number of corresponding characteristics.
In this embodiment, the corresponding features include discrete features and continuous features, the discrete features are subjected to one-hot encoding processing to convert the discrete features into binary features, and the binary features are combined into binary vectors to obtain first feature vectors of the discrete features. And carrying out standardization processing on the continuous features so that all numerical values in the feature vector of the continuous features are within a specific interval range, and obtaining a first feature vector of the continuous features. And merging the first eigenvector of the discrete features and the first eigenvector of the continuous features to obtain a first eigenvector matrix.
The discrete features are, for example, types of arrows on two sides of the current lane line, and the types of the lane line on two sides of the current lane line may be straight, left-handed, right-handed, u-turn, straight + left-handed, straight + right-handed, and the like, and if the types of the arrows on two sides of the current lane line have m values, the unique hot code can convert the values into m binary features (represented by 0 or 1), and then combine the m binary features into a binary vector, so that the corresponding feature of the types of the arrows on two sides of the current lane line can be expressed by the corresponding binary vector. For example, the feature vectors of the types of the arrows on both sides of the current lane line are obtained as {1,0, 0, 1, 1 }.
The continuous features are, for example, the number of lane lines on both sides of the current lane line, and the feature vectors of the continuous features may be normalized to obtain a first feature vector of the continuous features, where all values in the feature vectors fall within a specific interval range. For example, the feature vector of the number of lane lines on both sides of the current lane line is {5, 3, 4, 5, 4}, and the feature vector is scaled so that the feature vector is {1, 0.6, 0.8, 1, 0.8}, so as to avoid affecting the machine learning model parameters. Combining the characteristic vectors of the number of the lane lines on two sides of the current lane line with the characteristic vectors of the types of the arrows on two sides of the current lane line to obtain a first characteristic matrix of
Figure BDA0003194322360000141
S405, inputting the label vector of the training data set and the first feature matrix into the established machine learning model for training.
And simultaneously inputting the label vector of the training data set and the first feature matrix into the established machine learning model for training. After the machine learning model learns the feature vectors of the corresponding features and the corresponding labels of the corresponding features in the training data set, data without labels can be judged so as to identify lane grouping lines.
S406, collecting a test data set, wherein each lane line data in the test data set is provided with a corresponding second label, and the second label comprises a grouping line or a non-grouping line.
S407, extracting corresponding features from each lane line data in the test data set, wherein the corresponding features comprise: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line.
Step S406 is implemented in a manner similar to that of step S304, and step S407 is implemented in a manner similar to that of step S305, which is not described herein again.
S408, combining and representing the second labels of each lane line data in the test data set by using vectors, and obtaining label vectors in the test data set.
S409, obtaining a second feature vector of each corresponding feature in the test data set, and combining the second feature vectors to obtain a second feature matrix.
Step S408 is implemented in a manner similar to that of step S403, and step S409 is implemented in a manner similar to that of step S404, which are not described herein again.
S410, inputting the label vector of the test data set and the second feature matrix into a trained machine learning model for verification.
The machine learning model can select a Support Vector Machine (SVM) model, and the SVM model is used as a classical two-classification model and has strong advantages in the treatment of small samples, high-latitude characteristics and nonlinear problems. An SVM model initialized by all parameters may be obtained through a sklern library. And then, inputting the label vectors of the training set and the first feature matrix into a support vector machine model with initialized parameters for training. And then, inputting the label vector and the second feature matrix of the test data set into the trained support vector machine model, and outputting the prediction result of the test data set. And then, comparing the prediction result of the test data set with the label vector of the test data set, if the comparison result is greater than or equal to the preset precision, indicating that the trained SVM model meets the requirement, if the comparison result is less than the preset precision, adjusting the super-parameters of the model, such as a C parameter and a gamma parameter in an SVM algorithm, wherein the C parameter is a punishment item for limiting overfitting, the gamma parameter controls the width of the RBF core, and the gamma parameter is checked by adjusting the C parameter until the comparison result obtained after the prediction result of the test data set is compared with the label vector of the test data set reaches the preset precision.
S411, extracting lane line characteristics of a lane line to be detected, characteristics of road surface elements around the lane line to be detected and relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected.
S412, inputting the corresponding characteristics of the lane line to be detected into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises a lane grouping line.
In this embodiment, the lane line to be detected includes a lane line having a topological connection relationship with the lane grouping line, so that after the lane to be detected is input into the machine learning model to obtain the lane grouping line, the lane line to be detected may also be input into the graph model to obtain the lane line having a topological connection relationship with the lane grouping line through the graph model, and then the lane grouping line output by the machine learning model and the lane line having a topological connection relationship with the lane grouping line output by the graph model are taken as a final lane grouping line.
Before the lane line to be detected is input into the graph model, the end points of the known lane lines are used as vertexes, the graph data structure is established by taking the known lane lines as sides, the connected sides are obtained through the degree of each vertex, and the connection relation between the sides is established to obtain the graph model. The method comprises the steps that a lane line is known to have a topological connection relation with a lane grouping line, and a graph model is trained through the lane line having the topological connection relation with the lane grouping line, so that the graph model can reflect the topological connection relation of each lane line, and the lane line having the topological connection relation with the lane grouping line can be obtained after the lane line to be detected is input into the graph model.
Fig. 5 is a schematic structural diagram illustrating an apparatus for recognizing lane grouping lines according to an embodiment of the present invention, where the apparatus 10 for recognizing lane grouping lines according to the embodiment includes:
the extraction module 11 is configured to extract corresponding features of the lane line to be detected, where the corresponding features include: the lane line characteristics, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected;
and the control module 12 is configured to input the lane line characteristics of the lane line to be predicted, the characteristics of the road surface elements around the lane line to be predicted, and the relative characteristics of the lane line to be predicted and the road surface elements around the lane line to be predicted into a trained machine learning model, and output a prediction result, where the prediction result includes a lane grouping line.
The device 10 for identifying lane grouping lines provided in the embodiment of the present application may implement the above method embodiment, and specific implementation principles and technical effects thereof may refer to the above method embodiment, which is not described herein again.
Fig. 6 shows a hardware structure diagram of an electronic device according to an embodiment of the present application. As shown in fig. 6, the electronic device 20 is configured to implement the operations corresponding to the electronic device in any of the method embodiments described above, where the electronic device 20 of this embodiment may include: memory 21, processor 22 and communication interface 23.
A memory 21 for storing computer instructions.
A processor 22 for executing computer instructions stored by the memory to implement the method of identifying lane grouping lines in the above-described embodiments. Reference may be made in particular to the description relating to the method embodiments described above.
Alternatively, the memory 21 may be separate or integrated with the processor 22.
A communication interface 23, which may be connected to the processor 21.
The electronic device provided in this embodiment may be used to execute the above method for recognizing lane grouping lines, and its implementation manner and technical effect are similar, which are not described herein again.
The present application also provides a computer readable storage medium, in which computer instructions are stored, and the computer instructions are executed by a processor to implement the methods provided by the above-mentioned various embodiments.
The present application also provides a computer program product comprising computer instructions stored in a computer readable storage medium. The computer instructions may be read by at least one processor of the device from a computer-readable storage medium, and execution of the computer instructions by the at least one processor causes the device to perform the methods provided by the various embodiments described above.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same. Although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: it is also possible to modify the solutions described in the previous embodiments or to substitute some or all of them with equivalents. And the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present application.

Claims (14)

1. A method of identifying lane grouping lines, the method comprising:
extracting corresponding characteristics of the lane line to be detected, wherein the corresponding characteristics comprise: the lane line characteristics, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected;
and inputting the corresponding characteristics of the lane line to be detected into the trained machine learning model, and outputting a prediction result, wherein the prediction result comprises a lane grouping line.
2. The method of claim 1, wherein the lane marking features comprise: at least one of the mark, the geometric length, the direction, the type and the track distance of the lane line to be detected;
the characteristics of the road surface elements around the lane line to be measured include: the type of the arrow on the two sides of the lane line to be detected, and/or at least one of the number, the type, the width and the direction of the lane line on the two sides of the lane line to be detected;
the lane line characteristics and the relative characteristics of the road surface elements around the lane line to be detected include: the system comprises an adjacency matrix used for identifying the topological connection relation of the lane lines to be detected, and/or the distance between the lane lines to be detected and the nearest road surface identification, and/or at least one of the relative position relation, the width ratio, the height difference of an external matrix, the curvature difference and the angle difference of the lane lines to be detected on two sides of the lane lines to be detected and adjacent to the lane lines to be detected.
3. The method of claim 1 or 2, wherein prior to inputting the respective features of the lane line under test into the machine learning model, the method further comprises:
collecting a training data set, wherein each lane line data in the training data set is provided with a corresponding first label, and the first label comprises a grouping line or a non-grouping line;
extracting a respective feature from each lane line data in the training data set, the respective feature comprising: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line;
and inputting the first label and the corresponding features extracted from the training data set into a built machine learning model for training.
4. The method of claim 3, wherein prior to inputting the respective features of the lane line under test into the machine learning model, the method further comprises:
collecting a test data set, wherein each lane line data in the test data set is provided with a corresponding second label, and the second label comprises a grouping line or a non-grouping line;
extracting a respective feature from each lane line data in the test data set, the respective feature comprising: the lane line characteristics of the current lane line, the characteristics of the road surface elements around the current lane line, and the relative characteristics of the current lane line and the road surface elements around the current lane line;
and inputting the second label and the corresponding features extracted from the test data set into a trained machine learning model for verification.
5. The method of claim 4, wherein after the acquiring a training data set, the method further comprises:
combining and representing the first label of each lane line data in a training data set by using a vector to obtain a label vector of the training data set;
acquiring a first feature vector of each corresponding feature in the training data set, and merging the first feature vectors of the corresponding features of each lane line to obtain a first feature matrix;
correspondingly, the training of the machine learning model established by inputting the first label and the corresponding features extracted from the training data set specifically includes:
and inputting the label vector of the training data set and the first characteristic matrix into an established machine learning model for training.
6. The method of claim 5, wherein the respective features comprise discrete features and continuous features;
the obtaining of the first feature vector of each corresponding feature in the training data set, and merging the first feature vectors of the corresponding features of each lane line to obtain the first feature matrix specifically includes:
adopting one-hot encoding processing on the discrete features to convert the discrete features into binary features, and combining the binary features into a binary vector to obtain a first feature vector of the discrete features;
normalizing the continuous features to enable all numerical values in feature vectors of the continuous features to be within a specific interval range so as to obtain first feature vectors of the continuous features;
and combining the first feature vector of the discrete features and the first feature vector of the continuous features to obtain a first feature matrix.
7. The method of claim 5, wherein after said collecting a test data set, the method further comprises:
combining and representing second labels of each lane line data in a test data set by using vectors to obtain label vectors in the test data set;
acquiring a second eigenvector of each corresponding characteristic in the test data set, and merging the second eigenvectors of the corresponding characteristic of each lane line to obtain a second eigenvector matrix;
correspondingly, the inputting the second label and the corresponding features extracted from the test data set into the trained machine learning model for verification specifically includes:
and inputting the label vector of the test data set and the second feature matrix into a trained machine learning model for verification.
8. The method of claim 7, wherein the machine learning model comprises a support vector machine model;
inputting the label vector of the training set and the first feature matrix into a machine learning model established for training, and specifically comprising:
inputting the label vectors of the training set and the first feature matrix into a support vector machine model with initialized parameters for training;
inputting the label vector of the test data set and the second feature matrix into a trained machine learning model for verification, and specifically comprising:
inputting the label vector of the test data set and the second feature matrix into the trained support vector machine model, and outputting a prediction result of the test data set;
and comparing the prediction result of the test data set with the label vector of the test set.
9. The method of claim 8, wherein if the comparison of the predicted result of the test data set and the tag vector of the test data set is greater than or equal to a predetermined value, the support vector machine model is saved;
and if the comparison result of the prediction result of the test data set and the label vector of the test data set is smaller than a preset value, adjusting the hyper-parameters of the support vector machine model initialized by the parameters until the comparison result is larger than or equal to the preset value.
10. The method of claim 4, wherein the data of the training data set is larger than the data of the validation data set.
11. The method according to claim 1, wherein the lane line under test comprises a lane line having a topological relationship with a lane line grouping line;
after acquiring the lane grouping line, the method further comprises:
inputting the lane line to be detected into a graph model, and outputting a grouping line having a topological relation with the lane grouping line;
and taking the lane grouping line and the grouping line having the topological relation with the lane grouping line as a final lane grouping line.
12. The method of claim 11, wherein prior to inputting the lane line under test into the graphical model, the method further comprises:
and establishing graph model data by taking the end points of the known lane lines as vertexes and the known lane lines as sides, obtaining the connected sides through the degree of each vertex, and establishing the connection relation between the sides to obtain the graph model, wherein the known lane lines are the lane lines which are in topological connection with the lane grouping lines.
13. An apparatus for recognizing a lane grouping line, the apparatus comprising:
the extraction module is used for extracting corresponding characteristics of the lane line to be detected, and the corresponding characteristics comprise: the lane line characteristics, the characteristics of the road surface elements around the lane line to be detected and the relative characteristics of the lane line to be detected and the road surface elements around the lane line to be detected;
and the control module is used for inputting the corresponding characteristics of the lane line to be tested into the trained machine learning model and outputting a prediction result, wherein the prediction result comprises a lane grouping line.
14. An electronic device, characterized in that the device comprises: a memory, a processor;
the memory is to store computer instructions; the processor is configured to implement the method of identifying lane grouping lines of any of claims 1 to 12 in accordance with the computer instructions stored by the memory.
CN202110886471.7A 2021-08-03 2021-08-03 Method, device and equipment for identifying lane grouping lines Active CN113591730B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110886471.7A CN113591730B (en) 2021-08-03 2021-08-03 Method, device and equipment for identifying lane grouping lines

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110886471.7A CN113591730B (en) 2021-08-03 2021-08-03 Method, device and equipment for identifying lane grouping lines

Publications (2)

Publication Number Publication Date
CN113591730A true CN113591730A (en) 2021-11-02
CN113591730B CN113591730B (en) 2023-11-10

Family

ID=78254416

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110886471.7A Active CN113591730B (en) 2021-08-03 2021-08-03 Method, device and equipment for identifying lane grouping lines

Country Status (1)

Country Link
CN (1) CN113591730B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117746362A (en) * 2023-12-05 2024-03-22 北京卓视智通科技有限责任公司 Method and system for detecting continuous lane change of vehicle, storage medium and electronic equipment

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090268026A1 (en) * 2008-04-23 2009-10-29 Honda Motor Co., Ltd. Lane marker recognizing apparatus
CN106326822A (en) * 2015-07-07 2017-01-11 北京易车互联信息技术有限公司 Method and device for detecting lane line
CN109711341A (en) * 2018-12-27 2019-05-03 宽凳（北京）科技有限公司 A kind of virtual lane line recognition methods and device, equipment, medium
CN110390240A (en) * 2018-04-18 2019-10-29 百度（美国）有限责任公司 Lane post-processing in automatic driving vehicle
US20190329778A1 (en) * 2018-04-27 2019-10-31 Honda Motor Co., Ltd. Merge behavior systems and methods for merging vehicles
CN110704560A (en) * 2019-09-17 2020-01-17 武汉中海庭数据技术有限公司 Method and device for structuring lane line group based on road level topology
CN110796066A (en) * 2019-10-26 2020-02-14 武汉中海庭数据技术有限公司 Lane line group construction method and device
CN111316288A (en) * 2019-02-28 2020-06-19 深圳市大疆创新科技有限公司 Road structure information extraction method, unmanned aerial vehicle and automatic driving system
WO2020164010A1 (en) * 2019-02-13 2020-08-20 深圳市大疆创新科技有限公司 Lane line detection method, device, system, vehicle and storage medium
CN111563463A (en) * 2020-05-11 2020-08-21 上海眼控科技股份有限公司 Method and device for identifying road lane lines, electronic equipment and storage medium
CN112050821A (en) * 2020-09-11 2020-12-08 湖北亿咖通科技有限公司 Lane line polymerization method
CN112734927A (en) * 2021-03-31 2021-04-30 湖北亿咖通科技有限公司 Method and device for simplifying high-precision map lane line and computer storage medium
CN112825196A (en) * 2019-11-20 2021-05-21 阿里巴巴集团控股有限公司 Road marker determining method and device, storage medium and editing platform

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090268026A1 (en) * 2008-04-23 2009-10-29 Honda Motor Co., Ltd. Lane marker recognizing apparatus
CN106326822A (en) * 2015-07-07 2017-01-11 北京易车互联信息技术有限公司 Method and device for detecting lane line
CN110390240A (en) * 2018-04-18 2019-10-29 百度（美国）有限责任公司 Lane post-processing in automatic driving vehicle
US20190329778A1 (en) * 2018-04-27 2019-10-31 Honda Motor Co., Ltd. Merge behavior systems and methods for merging vehicles
CN109711341A (en) * 2018-12-27 2019-05-03 宽凳（北京）科技有限公司 A kind of virtual lane line recognition methods and device, equipment, medium
WO2020164010A1 (en) * 2019-02-13 2020-08-20 深圳市大疆创新科技有限公司 Lane line detection method, device, system, vehicle and storage medium
WO2020172875A1 (en) * 2019-02-28 2020-09-03 深圳市大疆创新科技有限公司 Method for extracting road structure information, unmanned aerial vehicle, and automatic driving system
CN111316288A (en) * 2019-02-28 2020-06-19 深圳市大疆创新科技有限公司 Road structure information extraction method, unmanned aerial vehicle and automatic driving system
CN110704560A (en) * 2019-09-17 2020-01-17 武汉中海庭数据技术有限公司 Method and device for structuring lane line group based on road level topology
CN110796066A (en) * 2019-10-26 2020-02-14 武汉中海庭数据技术有限公司 Lane line group construction method and device
CN112825196A (en) * 2019-11-20 2021-05-21 阿里巴巴集团控股有限公司 Road marker determining method and device, storage medium and editing platform
CN111563463A (en) * 2020-05-11 2020-08-21 上海眼控科技股份有限公司 Method and device for identifying road lane lines, electronic equipment and storage medium
CN112050821A (en) * 2020-09-11 2020-12-08 湖北亿咖通科技有限公司 Lane line polymerization method
CN112734927A (en) * 2021-03-31 2021-04-30 湖北亿咖通科技有限公司 Method and device for simplifying high-precision map lane line and computer storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117746362A (en) * 2023-12-05 2024-03-22 北京卓视智通科技有限责任公司 Method and system for detecting continuous lane change of vehicle, storage medium and electronic equipment

Also Published As

Publication number Publication date
CN113591730B (en) 2023-11-10

Similar Documents

Publication Publication Date Title
CN108229474B (en) Licence plate recognition method, device and electronic equipment
CN109558823B (en) Vehicle identification method and system for searching images by images
Wu et al. Automatic road extraction from high-resolution remote sensing images using a method based on densely connected spatial feature-enhanced pyramid
CN111680701B (en) Training method and device of image recognition model and image recognition method and device
CN113223068A (en) Multi-modal image registration method and system based on depth global features
CN112116950B (en) Protein folding identification method based on depth measurement learning
CN113591730A (en) Method, device and equipment for recognizing lane grouping line
CN112418037A (en) Method and system for identifying lane lines in satellite picture, electronic device and storage medium
CN114913386A (en) Training method of multi-target tracking model and multi-target tracking method
CN117475148A (en) Cargo segmentation method and system based on SAM and YOLOV8n
CN113238797A (en) Code feature extraction method and system based on hierarchical comparison learning
CN112052819A (en) Pedestrian re-identification method, device, equipment and storage medium
CN114612659A (en) Power equipment segmentation method and system based on fusion mode contrast learning
CN112084843B (en) Multispectral river channel remote sensing monitoring method based on semi-supervised learning
CN113673540A (en) Target detection method based on positioning information guidance
CN112381034A (en) Lane line detection method, device, equipment and storage medium
CN116977859A (en) Weak supervision target detection method based on multi-scale image cutting and instance difficulty
CN112464832A (en) Data processing method and intelligent device based on building map recognition and artificial intelligence
CN117058459A (en) Rapid pavement disease detection method and system based on YOLOV7 algorithm
CN109583584B (en) Method and system for enabling CNN with full connection layer to accept indefinite shape input
CN110705695A (en) Method, device, equipment and storage medium for searching model structure
CN115861306A (en) Industrial product abnormity detection method based on self-supervision jigsaw module
CN114067243A (en) Automatic driving scene recognition method, system, device and storage medium
CN114496068A (en) Protein secondary structure prediction method, device, equipment and storage medium
CN114495269A (en) Pedestrian re-identification method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20231227

Address after: Room 601, Building 1, No. 1782 Jiangling Road, Xixing Street, Binjiang District, Hangzhou City, Zhejiang Province, 310000

Patentee after: Hangzhou Langge Technology Co.,Ltd.

Address before: 430056 building B, building 7, Qidi Xiexin science and Innovation Park, South Taizi Lake innovation Valley, Wuhan Economic and Technological Development Zone, Wuhan City, Hubei Province (qdxx-f7b)

Patentee before: HUBEI ECARX TECHNOLOGY Co.,Ltd.