CN116664955A - Point cloud annotation type conversion method and device, terminal equipment and storage medium - Google Patents
Point cloud annotation type conversion method and device, terminal equipment and storage medium Download PDFInfo
- Publication number
- CN116664955A CN116664955A CN202310795189.7A CN202310795189A CN116664955A CN 116664955 A CN116664955 A CN 116664955A CN 202310795189 A CN202310795189 A CN 202310795189A CN 116664955 A CN116664955 A CN 116664955A
- Authority
- CN
- China
- Prior art keywords
- label
- point cloud
- semantic segmentation
- task
- labeling
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000006243 chemical reaction Methods 0.000 title claims abstract description 46
- 230000011218 segmentation Effects 0.000 claims abstract description 152
- 238000002372 labelling Methods 0.000 claims abstract description 132
- 238000001514 detection method Methods 0.000 claims abstract description 118
- 238000013507 mapping Methods 0.000 claims abstract description 42
- 238000004458 analytical method Methods 0.000 description 10
- 238000004891 communication Methods 0.000 description 8
- 230000008901 benefit Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000012790 confirmation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Processing Or Creating Images (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Document Processing Apparatus (AREA)
Abstract
The invention relates to the technical field of point cloud labeling, in particular to a point cloud labeling type conversion method, a device, terminal equipment and a storage medium, wherein the method comprises the following steps: acquiring a marked data set corresponding to a target detection task and task flow label configuration corresponding to a semantic segmentation task; acquiring preset label mapping, and determining semantic segmentation labels corresponding to target detection labels of each point cloud in the marked data set based on the label mapping; generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result. The method and the device can convert the annotation data of the point cloud target detection task into the annotation data of the semantic segmentation task, realize the conversion of the annotation type and improve the efficiency of point cloud annotation.
Description
Technical Field
The present invention relates to the field of point cloud labeling technologies, and in particular, to a method and an apparatus for converting a point cloud labeling type, a terminal device, and a storage medium.
Background
Point cloud semantic segmentation labeling is the process of assigning each point in a point cloud to a predefined semantic category. Point clouds are a collection of points sampled in three-dimensional space that have found wide application in the fields of computer vision and robotics because of their ability to capture the true shape and geometry of objects. The purpose of point cloud semantic segmentation is to understand objects in the point cloud and assign semantic tags to them. This is very useful for autopilot, such as: in an autonomous car, point cloud semantic segmentation may help identify roads, vehicles, pedestrians, and other obstacles, thereby helping the vehicle to plan and execute a safe driving route.
At present, a common point cloud semantic segmentation method generally converts 3d point cloud into manual voxel grid features or multi-view image features, and then sends the features into a deep learning network for feature extraction, so that the feature conversion method has large data size, complex calculation and low point cloud segmentation efficiency, and therefore, the traditional point cloud semantic segmentation method is low in efficiency when analyzing the real shape and the geometric structure of an object.
Accordingly, there is a need for improvement and advancement in the art.
Disclosure of Invention
The application aims to solve the technical problem that the prior art has low efficiency when analyzing the real shape and the geometric structure of an object by adopting the traditional point cloud semantic segmentation method. Is a problem of (a).
In order to solve the technical problems, the technical scheme adopted by the application is as follows:
in a first aspect, the present application provides a method for converting a point cloud label type, where the method includes:
acquiring a marked data set corresponding to a target detection task and task flow label configuration corresponding to a semantic segmentation task;
acquiring preset label mapping, and determining semantic segmentation labels corresponding to target detection labels of each point cloud in the marked data set based on the label mapping;
generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result.
According to the technical means, the labeling data of the point cloud target detection task are converted into the labeling data of the semantic segmentation task through the preset label mapping, and the time used by the point cloud target detection task when realizing the real shape and the geometric structure of the object is obviously lower than the time used by the point cloud semantic segmentation, so that the efficiency of point cloud target detection in the point cloud labeling is higher. Therefore, the embodiment of the application utilizes the advantage of high labeling efficiency of the point cloud target detection task, only converts the labeling data of the point cloud target detection task into the labeling data of the semantic segmentation task, and saves the time of point cloud labeling while meeting the data use requirement of the semantic segmentation task.
In one embodiment of the present application, the determining, based on the tag map, a semantic segmentation tag corresponding to a target detection tag of each point cloud in the labeled dataset includes:
based on the label mapping, the label mapping is used for reflecting the corresponding relation between a source detection label and a source semantic label of each point cloud in the target detection task, wherein the source detection label is a label determined according to the target detection task, and the source semantic label is a label determined according to the semantic segmentation task;
and matching the target detection label of each point cloud in the marked data set with the corresponding relation to obtain the semantic segmentation label.
According to the technical means, the label mapping is utilized to convert the label of the target detection task into the label of the semantic segmentation task, so that the label detection method is more accurate and quicker.
In one embodiment of the present application, the matching the target detection tag of each point cloud in the labeled dataset with the corresponding relationship to obtain the semantic segmentation tag includes:
acquiring label information of a target detection label of each point cloud in the marked data set, and determining a source detection label corresponding to the target detection label of each point cloud in the marked data set based on the label information;
And determining a source semantic tag corresponding to the source detection tag of each point cloud based on the corresponding relation, and taking the source semantic tag as the semantic segmentation tag.
According to the technical means, the embodiment of the application utilizes the corresponding relation set in the label mapping to match the label information of the target detection label of each point cloud in the marked data set, so that the semantic segmentation labels can be determined sequentially and rapidly, and the label conversion efficiency and accuracy are improved.
In one embodiment of the present application, the generating the labeling result of the semantic segmentation task based on the semantic segmentation tag and the task flow tag configuration includes:
acquiring a plurality of annotation frames corresponding to the semantic segmentation labels, analyzing point clouds in the annotated data set based on the annotation frames, and determining the relation between the point clouds and the annotation frames;
if the relation between the point cloud and the annotation frame is that the point cloud belongs to the annotation frame, associating the point cloud with the semantic segmentation label and generating a label index list;
and generating a labeling result of the semantic segmentation task based on the label index list and the task flow label configuration.
According to the technical means, the point cloud in the marked data set is analyzed, the tag index list is generated, the point cloud is associated with the semantic segmentation tag, and the marking result of the semantic segmentation task is obtained, so that the marking type of each point cloud in the marked data set can be converted, and the marking result of the semantic segmentation task can be obtained conveniently.
In one embodiment of the present application, the analyzing the point cloud in the marked data set based on the plurality of marking frames, determining the relationship between the point cloud and the marking frames includes:
determining convex hulls corresponding to the marking frames;
traversing the point cloud in the marked data set, and determining whether the point cloud belongs to any convex hull;
if the point cloud belongs to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud belongs to the annotation frame;
and if the point cloud does not belong to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud does not belong to the annotation frame.
According to the technical means, each point cloud in the marked data set is analyzed through the convex hull corresponding to the marking frame, whether the point cloud belongs to the marking frame is determined based on the convex hull, and further confirmation of the point cloud is achieved, so that marking results of each point cloud are guaranteed to meet the requirements of semantic segmentation tasks.
In one embodiment of the present application, the generating the labeling result of the semantic segmentation task based on the tag index list and the task flow tag configuration includes:
acquiring a tag format address in the task flow tag configuration;
and associating the semantic segmentation labels in the label index list with the label format addresses to obtain the labeling results.
According to the technical means, the embodiment of the application obtains the generation of the labeling result based on the association of the semantic segmentation label and the label format address, and completes the conversion of the labeling type.
In one embodiment of the present application, the issuing the point cloud annotation data set based on the annotation result includes:
determining an output path according to the label format address in the labeling result, and determining an output catalog corresponding to the output path;
and based on the output catalogue, a point cloud labeling data set corresponding to the labeling result is issued, wherein the point cloud labeling data set is a voice segmentation label corresponding to each point cloud of the semantic segmentation task.
According to the technical means, after the labeling result is obtained, the point cloud labeling data set is released, so that the point cloud labeling data set which is subjected to type conversion can be used when the semantic segmentation task is carried out, the point cloud labeling data set is converted according to the labeled data set corresponding to the target detection task, the advantage of high labeling efficiency of the point cloud target detection task is utilized, and the data use requirement of the point cloud semantic segmentation is met.
In a second aspect, an embodiment of the present application further provides a point cloud label type conversion device, where the device includes:
the data acquisition module is used for acquiring a marked data set corresponding to the target detection task and task flow label configuration corresponding to the semantic segmentation task;
the label conversion module is used for acquiring preset label mapping, and determining semantic segmentation labels corresponding to the target detection labels of each point cloud in the marked data set based on the label mapping;
the result generation module is used for generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result.
In one embodiment of the present application, the tag conversion module includes:
the relation determining unit is used for reflecting the corresponding relation between the source detection label and the source semantic label of each point cloud in the target detection task based on the label mapping, wherein the source detection label is a label determined according to the target detection task, and the source semantic label is a label determined according to the semantic segmentation task;
And the label matching unit is used for matching the target detection label of each point cloud in the marked data set with the corresponding relation to obtain the semantic segmentation label.
In one embodiment of the present application, the tag matching unit includes:
the first label mapping unit is used for acquiring label information of the target detection label of each point cloud in the marked data set, and determining a source detection label corresponding to the target detection label of each point cloud in the marked data set based on the label information;
the second label mapping unit is used for determining a source semantic label corresponding to the source detection label of each point cloud based on the corresponding relation, and taking the source semantic label as the semantic segmentation label.
In one embodiment of the present application, the result generation module includes:
the point cloud analysis unit is used for acquiring a plurality of marking frames corresponding to the semantic segmentation labels, analyzing the point cloud in the marked data set based on the marking frames and determining the relation between the point cloud and the marking frames;
the label index unit is used for associating the point cloud with the semantic segmentation label and generating a label index list if the relation between the point cloud and the label frame is that the point cloud belongs to the label frame;
And the result generating unit is used for generating a labeling result of the semantic segmentation task based on the label index list and the task flow label configuration.
In one embodiment of the present application, the point cloud analysis unit includes:
the convex hull determining unit is used for determining the convex hull corresponding to each labeling frame;
the point cloud judging unit is used for traversing the point cloud in the marked data set and determining whether the point cloud belongs to any convex hull or not;
the first analysis unit is used for determining that the point cloud belongs to the annotation frame if the point cloud belongs to any convex hull;
and the second analysis unit is used for determining that the relationship between the point cloud and the labeling frame is that the point cloud does not belong to the labeling frame if the point cloud does not belong to any convex hull.
In one embodiment of the present application, the result generation unit includes:
the address acquisition unit is used for acquiring the tag format address in the task flow tag configuration;
and the address association unit is used for associating the semantic segmentation labels in the label index list with the label format addresses to obtain the labeling results.
In one embodiment of the present application, the result generation module includes:
the catalog determining unit is used for determining an output path according to the label format address in the labeling result and determining an output catalog corresponding to the output path;
the data set issuing unit is used for issuing a point cloud labeling data set corresponding to the standard result based on the output catalog, wherein the point cloud labeling data set is a voice segmentation label corresponding to each point cloud of the semantic segmentation task.
In a third aspect, an embodiment of the present application further provides a terminal device, where the terminal device includes a memory, a processor, and a point cloud label type conversion program stored in the memory and capable of running on the processor, and when the processor executes the point cloud label type conversion program, the processor implements the steps of the point cloud label type conversion method in any one of the foregoing schemes.
In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium, where a point cloud label type conversion program is stored on the computer readable storage medium, where the point cloud label type conversion program, when executed by a processor, implements the steps of the point cloud label type conversion method according to any one of the above schemes.
The beneficial effects are that: compared with the prior art, the embodiment of the application provides a point cloud annotation type conversion method, the annotation data of a point cloud object detection task is converted into the annotation data of a semantic segmentation task through preset label mapping, and the time used by the point cloud object detection task when realizing the real shape and the geometric structure of an object is obviously lower than the time used when the point cloud semantic segmentation is carried out, so that the efficiency of point cloud object detection when the point cloud annotation is carried out is higher. Therefore, the embodiment of the application utilizes the advantage of high labeling efficiency of the point cloud target detection task, only converts the labeling data of the point cloud target detection task into the labeling data of the semantic segmentation task, and saves the time of point cloud labeling while meeting the data use requirement of the semantic segmentation task.
In addition, the embodiment of the application utilizes the corresponding relation set in the label mapping to match the label information of the target detection label of each point cloud in the marked data set, so that the semantic segmentation labels can be determined sequentially and rapidly, and the label conversion efficiency and accuracy are improved.
Additional aspects and advantages of the application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the application.
Drawings
FIG. 1 is a flowchart of a specific implementation of a method for converting a point cloud label type according to an embodiment of the present application;
fig. 2 is a business flow chart of a point cloud label type conversion method provided by an embodiment of the present application;
FIG. 3 is a type conversion flow chart of a point cloud label type conversion method provided by an embodiment of the application;
fig. 4 is a functional schematic diagram of a point cloud labeling type conversion device according to an embodiment of the present application;
fig. 5 is a schematic block diagram of a terminal device according to an embodiment of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are intended to illustrate the application and are not to be construed as limiting the application.
In the prior art, when analyzing the real shape and the geometric structure of an object, the traditional point cloud semantic segmentation method is low in efficiency. Therefore, the embodiment provides a point cloud annotation type conversion method, which can convert annotation data of a point cloud object detection task into annotation data of a semantic segmentation task based on the method, and the time used by the point cloud object detection task when realizing the real shape and the geometric structure of an object is obviously lower than the time used by the point cloud semantic segmentation, so that the efficiency of point cloud object detection when carrying out point cloud annotation is higher, and the efficiency of point cloud annotation can be improved. In specific application, the embodiment first obtains a labeled data set corresponding to the target detection task and task flow label configuration corresponding to the semantic segmentation task. And then, acquiring a preset label mapping, and determining a semantic segmentation label corresponding to the target detection label of each point cloud in the marked data set based on the label mapping. And after renting, generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result. Therefore, the method and the device for marking the point cloud target detection task have the advantages that the marking efficiency of the point cloud target detection task is high, only the marking data of the point cloud target detection task is required to be converted into the marking data of the semantic segmentation task, and the time of marking the point cloud is saved while the data using requirement of the semantic segmentation task is met.
The point cloud annotation type conversion method of the embodiment can be applied to terminal equipment, and the terminal equipment can be a vehicle-mounted controller, such as a vehicle-mounted central control computer. Or, the terminal device may also be a mobile terminal of the user, such as a mobile phone, where the mobile terminal may be connected to the vehicle-mounted terminal, so as to receive data transmitted by the vehicle-mounted terminal and perform corresponding analysis processing. Specifically, as shown in fig. 1, the method for converting a point cloud label type of the present embodiment includes the following steps:
step S100, a marked data set corresponding to the target detection task and task flow label configuration corresponding to the semantic segmentation task are obtained.
Because the embodiment is to convert the label data corresponding to the target detection task into the label data corresponding to the semantic segmentation task, the embodiment firstly obtains the labeled data set corresponding to the target detection task, and the labeled data set is the data set obtained after the target detection task has completed the data labeling, as shown in step 21 in fig. 2. Therefore, the present embodiment also acquires a task flow label configuration for generating the labeling result of the semantic segmentation task in the subsequent step, as shown in step 22 of fig. 2.
Step 200, acquiring a preset label mapping, and determining semantic segmentation labels corresponding to the target detection labels of each point cloud in the marked data set based on the label mapping.
The embodiment realizes the conversion of the labeling data between the target detection task and the semantic segmentation task based on label mapping. In this embodiment, the label mapping may reflect the correspondence between the target detection label and the semantic segmentation label, so that the semantic segmentation label corresponding to the target detection label of each point cloud in the labeled data set in the target detection task may be determined based on the label mapping, so as to implement conversion of the labeled data between the target detection task and the semantic segmentation task, as shown in step 23 in fig. 2, that is, the data processing task conversion format is performed.
In one implementation manner, when implementing label conversion, the embodiment includes the following steps:
step S201, based on the label mapping, the label mapping is used for reflecting the corresponding relation between a source detection label and a source semantic label of each point cloud in the target detection task, wherein the source detection label is a label determined according to the target detection task, and the source semantic label is a label determined according to the semantic segmentation task;
Step S202, matching the target detection label of each point cloud in the marked data set with the corresponding relation to obtain the semantic segmentation label.
When the method is applied specifically, when the label mapping is set, the source detection label is set for each point cloud according to the target detection task, and the source semantic label is set for each point cloud according to the semantic segmentation task. And setting the corresponding relation between the source detection label and the source semantic label to obtain the label mapping. The corresponding relation is the relation between the target detection task and the semantic segmentation task, and is the basis for realizing the conversion of the annotation data. Based on the correspondence, a label mapping is also established. In a specific application, the tag map of this embodiment may be stored in source Input File (source input file), and the storage path may be as follows: the data source is/mnt/bos/automatic/matrix-work space/matrix-process/task/task-1623303764691595266/data/data_source. When the label conversion is performed, firstly, the label information of the target detection label of each point cloud in the marked data set is acquired, wherein the label information can be information such as a label ID and the like for identifying the target detection label. And then, determining a source detection label corresponding to the target detection label of each point cloud in the marked data set based on the label information, namely, finding out which preset source detection label corresponds to the target detection label obtained in the target detection task. When the source detection tag is found, the embodiment can find the source semantic tag corresponding to the source detection tag of each point cloud from the corresponding relationship in the tag mapping, and the source semantic tag is the semantic segmentation tag in the embodiment. According to the method and the device, the corresponding relation set in the label mapping is utilized to match the label information of the target detection label of each point cloud in the marked data set, so that semantic segmentation labels can be determined sequentially and rapidly, and the label conversion efficiency and accuracy are improved.
And step S300, generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result.
After the semantic segmentation labels are determined, the embodiment can generate labeling results of the semantic segmentation tasks according to task flow label configuration. In the task flow label configuration in this embodiment, a label format address corresponding to a semantic segmentation label is set, where the label format address is preset in the task flow label configuration and can be used to generate a labeling result of the semantic segmentation task. After the labeling result is obtained, a point cloud labeling dataset corresponding to the semantic segmentation task may be issued, that is, a new dataset is generated according to the conversion result as shown in step 24 in fig. 2, so that conversion of labeling data between the target detection task and the semantic segmentation task is realized.
In a first aspect, the present embodiment includes the following steps when generating a labeling result:
step S301, a plurality of labeling frames corresponding to the semantic segmentation labels are obtained, point clouds in the labeled data set are analyzed based on the labeling frames, and the relation between the point clouds and the labeling frames is determined;
Step S302, if the relation between the point cloud and the labeling frame is that the point cloud belongs to the labeling frame, associating the point cloud with the semantic segmentation label, and generating a label index list;
and step S303, generating a labeling result of the semantic segmentation task based on the label index list and the task flow label configuration.
Specifically, since the principle of the object detection task and the semantic segmentation task are different, the object detection task is to detect an input image, and mark the position of the object on the image and which category the object belongs to on the position. The semantic segmentation task is to classify the input image pixel by pixel and mark out the object at the pixel level, and the labeling frames of the two are different. Therefore, in the embodiment, when generating the labeling result of the semantic segmentation task, analysis and screening are required to be performed on each point cloud in the labeled data set, so as to screen out the point clouds which do not belong to the semantic segmentation task. In specific application, the embodiment obtains a plurality of labeling frames corresponding to the semantic segmentation labels, and then analyzes the point cloud in the labeled data set based on the labeling frames to determine the relationship between the point cloud and the labeling frames.
Specifically, as shown in fig. 3, the present embodiment executes step 31 in fig. 3, namely, determines a convex hull corresponding to each labeling frame, where the convex hull is a polygon formed by the labeling frames, and then traverses the point cloud in the labeled dataset (i.e., step 32 in fig. 3) to determine whether the point cloud belongs to any convex hull (i.e., step 33 in fig. 3); and if the point cloud belongs to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud belongs to the annotation frame. And if the point cloud does not belong to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud does not belong to the annotation frame. Based on the above manner, each point cloud is analyzed, that is, step 35 in fig. 3 is performed to determine whether it is the last point cloud, and if it is not the last point cloud, step 32 is performed again until it is determined whether the last point cloud belongs to any convex hull. According to the method, each point cloud in the marked data set is analyzed through the convex hull corresponding to the marking frame, whether the point cloud belongs to the marking frame is determined based on the convex hull, and further confirmation of the point cloud is achieved, so that marking results of each point cloud are guaranteed to meet the requirements of semantic segmentation tasks.
When it is determined that the relationship between the point cloud and the labeling frame is that the point cloud belongs to the labeling frame, the embodiment may associate the point cloud with the semantic segmentation tag, and generate a tag index list (i.e. step 34 in fig. 3), that is, associate the semantic segmentation tag corresponding to the point cloud and add the semantic segmentation tag to a preset index empty list (e.g. index list), so as to obtain the tag index list. The tag index list of the embodiment can quickly find the point cloud corresponding to each semantic segmentation tag. Then, the embodiment may obtain the tag format address (i.e. json address) in the task flow tag configuration, and then associate the semantic segmentation tags in the tag index list with the tag format address, that is, execute step 36 in fig. 3, and construct json according to the semantic segmentation labeling format, so as to obtain the labeling result and the conversion of the labeling type.
After the labeling result is obtained, the embodiment can determine an output path according to the label format address in the labeling result, and determine an output directory corresponding to the output path. And then, based on the output catalogue, a point cloud labeling data set corresponding to the labeling result is issued, wherein the point cloud labeling data set is a voice segmentation label corresponding to each point cloud of the semantic segmentation task. When the method is specifically applied, when the point cloud annotation data set is released, the method can keep an output catalog with the annotated data set of the target detection task, a new file document needs to be written in a data_file set in mongab (mongab is extensible, high-performance, open source mode free and document oriented NoSQL), and a annotation result is written in the mongab according to a pre-annotated record form, as shown in step 25 in fig. 2, the release of the point cloud annotation data set can be completed by writing the new data set in the mongab. After the point cloud labeling data set is released, the point cloud labeling data set can be configured with labels and sent to a labeling system (as shown in step 26 in fig. 2, a new data set is sent to a label), so that a semantic segmentation result after semantic segmentation labeling can be seen on the labeling system, and then step 27 in fig. 2 is executed to perform a new round of labeling. Therefore, after the labeling result is obtained, the point cloud labeling data set is released, so that the point cloud labeling data set which is subjected to type conversion can be used when the semantic segmentation task is carried out, and the point cloud labeling data set is converted according to the labeled data set corresponding to the target detection task, so that the advantage of high labeling efficiency of the point cloud target detection task is utilized, and the data use requirement of the point cloud semantic segmentation is met.
Based on the above embodiment, the present application further provides a point cloud label type conversion device, as shown in fig. 4, where the point cloud label type conversion device 100 specifically includes: a data acquisition module 10, a tag conversion module 20 and a result generation module 30. Specifically, the data obtaining module 10 is configured to obtain a labeled data set corresponding to the target detection task and a task flow label configuration corresponding to the semantic segmentation task. The tag conversion module 20 is configured to obtain a preset tag mapping, and determine, based on the tag mapping, a semantic segmentation tag corresponding to a target detection tag of each point cloud in the labeled dataset. The result generating module 30 is configured to generate a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and issue a point cloud labeling dataset based on the labeling result.
In one embodiment of the present application, the tag conversion module includes:
the relation determining unit is used for reflecting the corresponding relation between the source detection label and the source semantic label of each point cloud in the target detection task based on the label mapping, wherein the source detection label is a label determined according to the target detection task, and the source semantic label is a label determined according to the semantic segmentation task;
And the label matching unit is used for matching the target detection label of each point cloud in the marked data set with the corresponding relation to obtain the semantic segmentation label.
In one embodiment of the present application, the tag matching unit includes:
the first label mapping unit is used for acquiring label information of the target detection label of each point cloud in the marked data set, and determining a source detection label corresponding to the target detection label of each point cloud in the marked data set based on the label information;
the second label mapping unit is used for determining a source semantic label corresponding to the source detection label of each point cloud based on the corresponding relation, and taking the source semantic label as the semantic segmentation label.
In one embodiment of the present application, the result generation module includes:
the point cloud analysis unit is used for acquiring a plurality of marking frames corresponding to the semantic segmentation labels, analyzing the point cloud in the marked data set based on the marking frames and determining the relation between the point cloud and the marking frames;
the label index unit is used for associating the point cloud with the semantic segmentation label and generating a label index list if the relation between the point cloud and the label frame is that the point cloud belongs to the label frame;
And the result generating unit is used for generating a labeling result of the semantic segmentation task based on the label index list and the task flow label configuration.
In one embodiment of the present application, the point cloud analysis unit includes:
the convex hull determining unit is used for determining the convex hull corresponding to each labeling frame;
the point cloud judging unit is used for traversing the point cloud in the marked data set and determining whether the point cloud belongs to any convex hull or not;
the first analysis unit is used for determining that the point cloud belongs to the annotation frame if the point cloud belongs to any convex hull;
and the second analysis unit is used for determining that the relationship between the point cloud and the labeling frame is that the point cloud does not belong to the labeling frame if the point cloud does not belong to any convex hull.
In one embodiment of the present application, the result generation unit includes:
the address acquisition unit is used for acquiring the tag format address in the task flow tag configuration;
and the address association unit is used for associating the semantic segmentation labels in the label index list with the label format addresses to obtain the labeling results.
In one embodiment of the present application, the result generation module includes:
the catalog determining unit is used for determining an output path according to the label format address in the labeling result and determining an output catalog corresponding to the output path;
the data set issuing unit is used for issuing a point cloud labeling data set corresponding to the standard result based on the output catalog, wherein the point cloud labeling data set is a voice segmentation label corresponding to each point cloud of the semantic segmentation task.
The working principle of each module in the point cloud labeling type conversion device 100 of the present embodiment is the same as that of each step in the above method embodiment, and will not be described here again.
The point cloud labeling type conversion device 100 of the embodiment converts labeling data of a point cloud target detection task into labeling data of a semantic segmentation task through preset label mapping, and when the point cloud target detection task achieves the real shape and the geometric structure of an object, the time used for the point cloud target detection task is obviously lower than the time used for the point cloud semantic segmentation, so that the efficiency of point cloud target detection in the point cloud labeling is higher. Therefore, the embodiment of the application utilizes the advantage of high labeling efficiency of the point cloud target detection task, only converts the labeling data of the point cloud target detection task into the labeling data of the semantic segmentation task, and saves the time of point cloud labeling while meeting the data use requirement of the semantic segmentation task.
Fig. 5 is a schematic structural diagram of a terminal device according to an embodiment of the present application. The terminal device may include: memory 501, processor 502, and a computer program stored on memory 501 and executable on processor 502. The processor 502 implements the point cloud label type conversion method provided in the above embodiment when executing a program.
Further, the terminal device further includes:
a communication interface 503 for communication between the memory 501 and the processor 502.
Memory 501 for storing a computer program executable on processor 502.
The memory 501 may include high-speed RAM memory and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
If the memory 501, the processor 502, and the communication interface 503 are implemented independently, the communication interface 503, the memory 501, and the processor 502 may be connected to each other via a bus and perform communication with each other. The bus may be an industry standard architecture (Industry Standard Architecture, abbreviated ISA) bus, an external device interconnect (Periphera l Component, abbreviated PCI) bus, or an extended industry standard architecture (Extended Industry Standard Architecture, abbreviated EISA) bus, among others. The buses may be divided into address buses, data buses, control buses, etc. For ease of illustration, only one thick line is shown in fig. 5, but not only one bus or one type of bus.
In a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 may perform communication with each other through internal interfaces. The processor 502 may be a central processing unit (Central Processing Unit, abbreviated as CPU) or an application specific integrated circuit (Application Specific Integrated Circuit, abbreviated as ASIC) or one or more integrated circuits configured to implement embodiments of the present application.
The present embodiment also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the point cloud annotation type conversion method as above.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "particular embodiments," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or N embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, "N" means at least two, for example, two, three, etc., unless specifically defined otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and additional implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order from that shown or discussed, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present application.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or N wires, a portable computer cartridge (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the N steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As with the other embodiments, if implemented in hardware, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.
In addition, each functional unit in the embodiments of the present application may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product.
The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like. While embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the application, and that variations, modifications, alternatives, and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the application.
Claims (10)
1. The point cloud annotation type conversion method is characterized by comprising the following steps of:
acquiring a marked data set corresponding to a target detection task and task flow label configuration corresponding to a semantic segmentation task;
acquiring preset label mapping, and determining semantic segmentation labels corresponding to target detection labels of each point cloud in the marked data set based on the label mapping;
generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result.
2. The method of claim 1, wherein determining, based on the tag map, a semantic segmentation tag corresponding to a target detection tag of each point cloud in the labeled dataset comprises:
Based on the label mapping, the label mapping is used for reflecting the corresponding relation between a source detection label and a source semantic label of each point cloud in the target detection task, wherein the source detection label is a label determined according to the target detection task, and the source semantic label is a label determined according to the semantic segmentation task;
and matching the target detection label of each point cloud in the marked data set with the corresponding relation to obtain the semantic segmentation label.
3. The method for converting a point cloud label type according to claim 2, wherein the matching the target detection label of each point cloud in the labeled dataset with the correspondence to obtain the semantic segmentation label includes:
acquiring label information of a target detection label of each point cloud in the marked data set, and determining a source detection label corresponding to the target detection label of each point cloud in the marked data set based on the label information;
and determining a source semantic tag corresponding to the source detection tag of each point cloud based on the corresponding relation, and taking the source semantic tag as the semantic segmentation tag.
4. The method of claim 1, wherein the generating the labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration comprises:
acquiring a plurality of annotation frames corresponding to the semantic segmentation labels, analyzing point clouds in the annotated data set based on the annotation frames, and determining the relation between the point clouds and the annotation frames;
if the relation between the point cloud and the annotation frame is that the point cloud belongs to the annotation frame, associating the point cloud with the semantic segmentation label and generating a label index list;
and generating a labeling result of the semantic segmentation task based on the label index list and the task flow label configuration.
5. The method of claim 4, wherein the analyzing the point clouds in the marked dataset based on the plurality of marking frames to determine the relationship between the point clouds and the marking frames comprises:
determining convex hulls corresponding to the marking frames;
traversing the point cloud in the marked data set, and determining whether the point cloud belongs to any convex hull;
If the point cloud belongs to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud belongs to the annotation frame;
and if the point cloud does not belong to any convex hull, determining that the relation between the point cloud and the annotation frame is that the point cloud does not belong to the annotation frame.
6. The method of claim 5, wherein generating the labeling result of the semantic segmentation task based on the tag index list and the task flow tag configuration comprises:
acquiring a tag format address in the task flow tag configuration;
and associating the semantic segmentation labels in the label index list with the label format addresses to obtain the labeling results.
7. The method of claim 6, wherein the issuing a point cloud annotation dataset based on the annotation result comprises:
determining an output path according to the label format address in the labeling result, and determining an output catalog corresponding to the output path;
and based on the output catalogue, a point cloud labeling data set corresponding to the labeling result is issued, wherein the point cloud labeling data set is a voice segmentation label corresponding to each point cloud of the semantic segmentation task.
8. A point cloud annotation type conversion device, the device comprising:
the data acquisition module is used for acquiring a marked data set corresponding to the target detection task and task flow label configuration corresponding to the semantic segmentation task;
the label conversion module is used for acquiring preset label mapping, and determining semantic segmentation labels corresponding to the target detection labels of each point cloud in the marked data set based on the label mapping;
the result generation module is used for generating a labeling result of the semantic segmentation task based on the semantic segmentation label and the task flow label configuration, and distributing a point cloud labeling data set based on the labeling result.
9. A terminal device, characterized in that the terminal device comprises a memory, a processor and a point cloud label type conversion program stored in the memory and operable on the processor, the processor implementing the steps of the point cloud label type conversion method according to any of claims 1-7 when executing the point cloud label type conversion program.
10. A computer readable storage medium, wherein a point cloud label type conversion program is stored on the computer readable storage medium, and when the point cloud label type conversion program is executed by a processor, the steps of the point cloud label type conversion method according to any one of claims 1-7 are implemented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310795189.7A CN116664955A (en) | 2023-06-30 | 2023-06-30 | Point cloud annotation type conversion method and device, terminal equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310795189.7A CN116664955A (en) | 2023-06-30 | 2023-06-30 | Point cloud annotation type conversion method and device, terminal equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116664955A true CN116664955A (en) | 2023-08-29 |
Family
ID=87719132
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310795189.7A Pending CN116664955A (en) | 2023-06-30 | 2023-06-30 | Point cloud annotation type conversion method and device, terminal equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116664955A (en) |
-
2023
- 2023-06-30 CN CN202310795189.7A patent/CN116664955A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3620966A1 (en) | Object detection method and apparatus for object detection | |
| CN109308681B (en) | Image processing method and device | |
| US11113546B2 (en) | Lane line processing method and device | |
| US20220122260A1 (en) | Method and apparatus for labeling point cloud data, electronic device, and computer-readable storage medium | |
| CN110674788A (en) | Vehicle damage assessment method and device | |
| EP3620945A1 (en) | Obstacle distribution simulation method, device and terminal based on multiple models | |
| US11256965B2 (en) | Apparatus and method for recognizing object using image | |
| CN112258519B (en) | Automatic extraction method and device for way-giving line of road in high-precision map making | |
| CN111402413B (en) | Three-dimensional visual positioning method and device, computing equipment and storage medium | |
| CN110084230B (en) | Image-based vehicle body direction detection method and device | |
| CN114820679B (en) | Image labeling method and device electronic device and storage medium | |
| CN111199087A (en) | Scene recognition method and device | |
| US11354461B2 (en) | Method and device for simulating a distribution of obstacles | |
| CN112990200A (en) | Data labeling method and device, computer equipment and storage medium | |
| CN113246858B (en) | Vehicle running state image generation method, device and system | |
| CN109300322B (en) | Guideline drawing method, apparatus, device, and medium | |
| CN114693963A (en) | Recognition model training and recognition method and device based on electric power data feature extraction | |
| CN112654999B (en) | Method and device for determining labeling information | |
| CN116664955A (en) | Point cloud annotation type conversion method and device, terminal equipment and storage medium | |
| CN113345101B (en) | Three-dimensional point cloud labeling method, device, equipment and storage medium | |
| CN116642490A (en) | Visual positioning navigation method based on hybrid map, robot and storage medium | |
| CN112946603B (en) | Road maintenance detection system based on laser radar and detection method thereof | |
| US20210207964A1 (en) | Verification Method And Device For Modeling Route, Unmanned Vehicle, And Storage Medium | |
| CN112132031B (en) | Vehicle style identification method and device, electronic equipment and storage medium | |
| CN115393379A (en) | Data annotation method and related product |
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 |