Web-based continuous frame point cloud data labeling method and device

Abstract

The application discloses a web-based continuous frame point cloud data labeling method and device, and the method comprises the following steps: obtaining a labeling request, responding to the labeling request, downloading continuous frame point cloud data to be labeled and corresponding 2D image data and storing the same in a local database; obtaining an initial labeling box of an object in a frame of point cloud data in the continuous frame point cloud data, automatically calculating a final labeling box based on the initial labeling box; and mapping the final labeling box to the 2D image data. The application can realize multi-frame big data loading and automatic convergence of the labeling box, improve labeling efficiency and reduce labeling difficulty, so that labeling personnel can complete continuous big data labeling of a specific scene in continuous time, improve data labeling accuracy and accelerate production of labeled data.

Description

Technical Field

[0001] This invention relates generally to the field of autonomous driving, and more specifically to a web-based method and apparatus for labeling continuous frame point cloud data. Background Technology

[0002] With the healthy development of the autonomous driving industry, various regions are continuously opening up autonomous driving test roads and improving relevant regulations, laying a solid foundation for the development of intelligent vehicles in China. The core modules of autonomous driving are the perception layer, decision-making layer, and control layer, primarily addressing the core questions of autonomous driving: Where am I? Where am I going? How do I get there? Obstacle detection is a crucial component of the perception layer and a prerequisite for achieving autonomous driving functions. Utilizing various sensors to acquire information about the vehicle's environment and surrounding vehicles, pedestrians, traffic lights, and road signs, capturing 2D visual data through cameras and 3D position data through LiDAR and millimeter-wave radar, provides data support for the vehicle's comprehensive decision-making and is the core measure for solving the "Where am I?" question.

[0003] While current artificial intelligence (AI) is relatively mature in terms of computing power and algorithms, addressing specific industry pain points through algorithm and application implementation requires collecting massive amounts of autonomous driving scenario data. This data needs to be processed and manually labeled to support algorithm training. In the field of autonomous driving annotation, a common method is to fuse single-frame 2D and 3D data from the web, simultaneously annotating images and point cloud data collected by both 2D (cameras) and 3D sensors (radar). However, this single-frame small point cloud annotation scheme is insufficient for algorithm training in all autonomous driving scenarios. It is necessary to establish connections between frames and support large point cloud annotation methods from the web to build training datasets and improve the capabilities of algorithms such as target tracking and semantic segmentation.

[0004] Therefore, in order to solve the above problems, a new web-based continuous frame point cloud data annotation method and device are needed to address the issues existing in current point cloud annotation schemes. Summary of the Invention

[0005] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. The summary section of this invention is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0006] According to one aspect of the present invention, a web-based continuous frame point cloud data annotation method is provided, the method comprising: obtaining an annotation request; in response to the annotation request, downloading continuous frame point cloud data to be annotated and corresponding 2D image data and storing them in a local database; obtaining an initial bounding box for an object in one frame of point cloud data in the continuous frame point cloud data; automatically calculating a final bounding box based on the initial bounding box; and mapping the final bounding box to the 2D image data.

[0007] In one embodiment, the method further includes: downloading continuous frame point cloud data and 2D image data to be labeled and obtaining the initial label box simultaneously using different processing resources.

[0008] In one embodiment, the local database is the browser's local database.

[0009] In one embodiment, automatically calculating the final annotation box based on the initial annotation box includes: obtaining the point cloud coordinates of the point cloud set within the initial annotation box; and calculating the minimum bounding rectangle of the object based on the point cloud coordinates, and automatically modifying the initial annotation box according to the minimum bounding rectangle to obtain the final annotation box.

[0010] In one embodiment, calculating the minimum bounding rectangle of the object based on the point cloud coordinates includes: estimating the height of the object based on the point cloud coordinates, and calculating an updated point cloud set within the initial bounding box based on the estimated height of the object; calculating an image representation of the object using the updated point cloud set; and calculating the size of the minimum bounding rectangle based on the image representation of the object, thereby obtaining the minimum bounding rectangle.

[0011] In one embodiment, the method further includes: transforming the point cloud coordinates from a radar coordinate system to a point cloud coordinate system before estimating the height of the object based on the point cloud coordinates.

[0012] In one embodiment, estimating the height of the object based on the point cloud coordinates includes: calculating the height difference between the highest and lowest points in the point cloud set; and estimating the height of the object by removing the ground height from the height difference, wherein calculating an updated point cloud set within the initial bounding box based on the estimated height of the object includes: removing points outside the estimated height range of the object from the point cloud set to obtain the updated point cloud set.

[0013] In one embodiment, calculating an image representation of the object using the updated point cloud set includes: transforming the updated point cloud set to an image coordinate system and calculating an image representation of the object based on the transformed point cloud set.

[0014] In one embodiment, the method further includes: after labeling the continuous frame point cloud data, performing multi-frame fitting playback on the labeling results.

[0015] According to another aspect of the present invention, a web-based continuous frame point cloud data annotation apparatus is provided. The apparatus includes a memory and a processor. The memory stores a computer program that is executed by the processor. When the computer program is executed by the processor, it causes the processor to perform the web-based continuous frame point cloud data annotation method as described above.

[0016] According to another aspect of the present invention, a computer-readable medium is provided, wherein computer-executable instructions are stored thereon, which, when executed, perform the web-based continuous frame point cloud data annotation method as described above.

[0017] The web-based continuous frame point cloud data annotation method and apparatus according to embodiments of the present invention can realize multi-frame big data loading and automatic convergence of annotation boxes, improve annotation efficiency, reduce annotation difficulty, enable annotators to complete continuous big data annotation of specific scenarios in a continuous time, improve data annotation accuracy, and accelerate the production of annotated data. Attached Figure Description

[0018] The following drawings, which are incorporated herein by reference as part of this invention, are provided for understanding the invention. The drawings illustrate embodiments of the invention and their descriptions, serving to explain the principles of the invention.

[0019] In the attached image:

[0020] Figure 1 An exemplary flowchart of a web-based continuous frame point cloud data annotation method according to an embodiment of the present invention is shown.

[0021] Figure 2 A schematic diagram illustrating an example image representation according to an embodiment of the present invention is shown; and

[0022] Figure 3 A schematic structural block diagram of a web-based continuous frame point cloud data annotation device according to an embodiment of the present invention is shown. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0024] As mentioned above, existing 2D / 3D fusion annotation is based on point cloud data and 2D image data collected by radar and cameras. The data is divided into a set according to the frame. After the annotation client pulls the data, it loads the point cloud data and 2D image data respectively. On the annotation page or client, the point cloud is annotated by drawing boxes and adding labels. The annotation boxes are mapped to the 2D image by transforming the radar coordinate system and the camera coordinate system to assist in accurate annotation.

[0025] Specifically, the data annotation process for 2D / 3D fusion annotation can generally be divided into the following steps:

[0026] Step 1: Data loading, retrieving the fused annotation data content and annotation parameters;

[0027] Step 2: Load and render the point cloud annotation area;

[0028] Step 3: Load 2D image information and display the image data captured by each camera;

[0029] Step 4: 3D point cloud bounding box annotation. By transforming the radar coordinate system and camera coordinate system, the drawn annotation box is projected onto the 2D image.

[0030] Step 5: Save the annotation results and export them to provide to the autonomous driving model engine for training.

[0031] Existing 2D / 3D fusion annotation schemes mainly involve extracting keyframes from the original data after point cloud parsing and video frame extraction and time sequence alignment, and then annotating each frame. Although this relatively simple method achieves data fusion annotation for autonomous driving, the lack of connection between individual frames makes it difficult to judge the annotation of the same target object. For scenarios such as target tracking and semantic segmentation, there is a lack of ability to load and annotate reconstructed point clouds and continuous frame point clouds. On the one hand, the data accuracy is insufficient, and on the other hand, it increases the difficulty and cost of data annotation.

[0032] In summary, current point cloud annotation methods for web applications have the following problems:

[0033] 1. The single-frame annotation method results in no connection between frames. The annotation process requires constantly switching point clouds to determine whether the annotated objects are the same objects, resulting in low annotation efficiency.

[0034] 2. Switching back and forth between single-frame point clouds results in low data loading efficiency.

[0035] 3. Given the complexity of urban road scenarios, the varying orientations of vehicles, and their high density, manual data annotation is currently costly.

[0036] 4. Since the point cloud data within a unit of time is temporally correlated, the current single-frame annotation method for extracting keyframes cannot guarantee the accuracy and quality of the data.

[0037] Therefore, in order to solve the above-mentioned problems of existing point cloud data annotation methods, the present invention provides a web-based continuous frame point cloud data annotation method, the method comprising: obtaining an annotation request; in response to the annotation request, downloading continuous frame point cloud data to be annotated and corresponding 2D image data and storing them in a local database; obtaining an initial annotation box for an object in one frame of point cloud data in the continuous frame point cloud data; automatically calculating a final annotation box based on the initial annotation box; and mapping the final annotation box to the 2D image data.

[0038] The web-based continuous frame point cloud data annotation method of the present invention can realize multi-frame big data loading and automatic convergence of annotation boxes, improve annotation efficiency, reduce annotation difficulty, enable annotators to complete continuous big data annotation of specific scenarios in a continuous time, improve data annotation accuracy, and accelerate the production of annotated data.

[0039] The following describes in detail, with reference to specific embodiments, the web-based continuous frame point cloud data annotation method and apparatus according to the present invention.

[0040] First, refer to Figure 1 , Figure 1 An exemplary flowchart of a web-based continuous frame point cloud data annotation method 100 according to an embodiment of the present invention is shown.

[0041] like Figure 1 As shown, the web-based continuous frame point cloud data annotation method 100 may include the following exemplary steps:

[0042] In step S110, an annotation request is obtained, and in response to the annotation request, the continuous frame point cloud data to be annotated and the corresponding 2D image data are downloaded and stored in the local database;

[0043] In step S120, an initial bounding box for an object in one frame of point cloud data from a series of consecutive frames is obtained, and the final bounding box is automatically calculated based on the initial bounding box; and

[0044] In step S130, the final bounding box is mapped to 2D image data.

[0045] In one embodiment, a labeling request can be obtained by a user clicking on a labeling task.

[0046] In real-world annotation scenarios, loading large amounts of point cloud and 2D image data on the web is extremely resource-intensive, and waiting for all resources to load takes too long. Therefore, in one embodiment, method 100 may further include: downloading the continuous frame point cloud data and 2D image data to be annotated and obtaining the initial annotation boxes using different processing resources simultaneously. In one embodiment, HTML5 Web Workers can be used to create a multi-threaded environment for JavaScript, allowing the main thread to create worker threads. The time-consuming operations of loading and parsing the remaining point cloud in the main thread can be distributed to the worker threads. Thus, after loading the first frame of point cloud, the main thread can begin processing annotation and rendering tasks, while the worker threads load and parse the remaining point cloud, without interference. After the worker threads complete their processing, the results are returned to the main thread via a message mechanism. This invention fully utilizes the advantages of multi-threading; loading large amounts of data has no impact on the annotation process, and it can handle rendering, user clicks, and other operations normally, ensuring that the main thread is not blocked and the annotation and rendering pages do not lag.

[0047] Because consecutive frames require batches of point cloud data, and a single point cloud contains hundreds of thousands of points, batch storage is very memory-intensive. Therefore, in one embodiment, the local database can be a browser-local database suitable for large data storage, such as IndexedDB. Compared to cookies and local storage, IndexedDB has the following characteristics:

[0048] a) Asynchronous: IndexedDB operations do not lock the browser, allowing users to perform other operations as well. This contrasts with local storage, which operates synchronously. Using IndexedDB to store data does not impact browser performance.

[0049] b) Large storage space: IndexedDB's storage space is much larger than Local Storage, generally no less than 250MB, and there is no upper limit.

[0050] c) Supports binary storage: IndexedDB can store not only strings, but also binary data (e.g., ArrayBuffer objects and Blob objects).

[0051] This invention stores the retrieved point cloud and image data in IndexedDB. When the page is refreshed or reloaded for the second time, the data cached in IndexedDB can be read directly, reducing the loading time caused by the second network request and improving the annotation efficiency after the second load.

[0052] In one embodiment, the step S120 of automatically calculating the final annotation box based on the initial annotation box may include the following steps:

[0053] Obtain the point cloud coordinates of the point cloud set within the initial annotation box; and

[0054] The minimum bounding rectangle of the object is calculated based on the point cloud coordinates, and the initial annotation box is automatically modified according to the minimum bounding rectangle to obtain the final annotation box.

[0055] Since the point cloud data is acquired by radar (e.g., lidar, millimeter-wave radar, etc.), the point cloud coordinates of the point cloud set within the initial annotation box are the coordinates of the radar coordinate system.

[0056] In one embodiment, calculating the minimum bounding rectangle of an object based on point cloud coordinates may include the following steps:

[0057] The height of the object is estimated based on the point cloud coordinates, and the updated point cloud set within the initial bounding box is calculated based on the estimated height of the object.

[0058] Compute an image representation of the object using the updated point cloud set; and

[0059] The object-based image representation calculates the size of the minimum bounding rectangle, thus obtaining the minimum bounding rectangle.

[0060] To facilitate calculation, in one embodiment, the point cloud coordinates are transformed from the radar coordinate system to the point cloud coordinate system before estimating the height of the object based on the point cloud coordinates. Specifically, the center point of the initial annotation box can be selected as the origin, and the coordinates of each point in the point cloud relative to the center point can be calculated, thereby transforming the point cloud coordinates from the radar coordinate system to the point cloud coordinate system.

[0061] In one embodiment, estimating the height of the object based on point cloud coordinates may include: calculating the height difference between the highest and lowest points in the point cloud set; and estimating the object's height by removing the ground height from the height difference. Since the ground is relatively low, it can be empirically determined which height ranges constitute the ground.

[0062] In one embodiment, calculating an updated point cloud set within an initial bounding box based on the estimated height of the object may include: removing points outside the estimated height range of the object from the point cloud set to obtain the updated point cloud set.

[0063] In one embodiment, computing an image representation of an object using an updated point cloud set may include: transforming the updated point cloud set to an image coordinate system, and computing the image representation of the object based on the transformed point cloud set. After transforming the point cloud set to an image coordinate system, the point cloud can be converted into an image representation dst using, for example, OpenCV's adaptiveThreshold function. Figure 2 An example image representing a schematic diagram of dst is shown according to an embodiment of the present invention.

[0064] Therefore, after obtaining the initial annotation box, the present invention can use the above-mentioned automatic fitting algorithm to achieve automatic convergence of the annotation box, thereby improving annotation efficiency and accuracy.

[0065] After the point cloud data is labeled, the final label box can be automatically mapped to the 2D image data, thereby helping to view the alignment of the point cloud on the 2D image.

[0066] Since the annotation is done in the point cloud coordinate system, and the points in the annotation box belong to the BBox coordinate system, the coordinates of the annotation box need to be transformed to the 2D image coordinate system as follows:

[0067] a) Bbox coordinates → Radar coordinates

[0068]

[0069] b) Radar coordinates → Camera coordinates

[0070]

[0071] c) Camera coordinates → Image coordinates

[0072] P I =kP C

[0073] In this diagram, subscripts L, B, and C represent radar, point cloud, and camera, respectively; P represents coordinates, R represents the rotation matrix, T represents the translation vector, and k represents the camera matrix. This represents the rotation matrix from the point cloud to the radar. This represents the rotation matrix from the radar to the camera. This represents the translation vector from the point cloud to the radar. This represents the translation vector from the radar to the camera.

[0074] After three transformations, we can obtain a 4x4 homogeneous coordinate matrix of the image as follows:

[0075]

[0076] By using homogeneous coordinate transformation to project the calculated matrix perspective onto the 2D image, the mapping of the 3D point cloud annotation box to the pixel position of the 2D image is completed.

[0077] In one embodiment, method 100 may further include: after labeling continuous frame point cloud data, performing multi-frame fitting playback on the labeling results. By smoothly switching between frames, the dynamic changes of point cloud and camera during object movement are simulated, supporting the switching of multiple sensors, facilitating dynamic tracking of the same object from various angles during the labeling process, and improving the accuracy and efficiency of labeling.

[0078] The web-based continuous frame point cloud data annotation method of the present invention can realize multi-frame big data loading and automatic convergence of annotation boxes, improve annotation efficiency, reduce annotation difficulty, enable annotators to complete continuous big data annotation of specific scenarios in a continuous time, improve data annotation accuracy, and accelerate the production of annotated data; and can also perform data validity checks through multi-frame fitting and playback, breaking through the point cloud data quality barrier.

[0079] This invention also provides a web-based continuous frame point cloud data annotation device 300. (Reference) Figure 3 , Figure 3 A schematic structural block diagram of a web-based continuous frame point cloud data annotation device 300 according to an embodiment of the present invention is shown. Figure 3 As shown, the web-based continuous frame point cloud data annotation device 300 may include a memory 310 and a processor 320. The memory 310 stores a computer program executed by the processor 320. When the computer program is executed by the processor 320, it causes the processor 320 to perform the web-based continuous frame point cloud data annotation method 100 described above according to an embodiment of the present invention. Those skilled in the art can understand the specific operation of the web-based continuous frame point cloud data annotation device 300 according to an embodiment of the present invention in conjunction with the foregoing description; for the sake of brevity, it will not be described in detail here.

[0080] The web-based continuous frame point cloud data annotation device of the present invention can realize multi-frame big data loading and automatic convergence of annotation boxes, improve annotation efficiency, reduce annotation difficulty, enable annotators to complete continuous big data annotation of specific scenarios in a continuous time, improve data annotation accuracy, and accelerate the production of annotation data; and can also perform data validity checks through multi-frame fitting playback, breaking through the point cloud data quality barrier.

[0081] The present invention also provides a computer-readable medium storing computer-executable instructions that, when executed, perform the corresponding steps of the web-based continuous frame point cloud data annotation method 100 as described above. Any tangible, non-transitory computer-readable medium may be used, including magnetic storage devices (hard disks, floppy disks, etc.), optical storage devices (CD-ROMs, DVDs, Blu-ray discs, etc.), flash memory, and / or the like. These computer-executable instructions may be loaded onto a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to form a machine, such that the instructions, which execute on the computer or other programmable data processing apparatus, can generate means for implementing a specified function. These computer-executable instructions may also be stored in a computer-readable storage medium that can instruct the computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable storage medium can form an article of manufacture including means for implementing the specified function. The computer-executable instructions may also be loaded onto a computer or other programmable data processing apparatus to perform a series of operational steps on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, can provide steps for implementing the specified function.

[0082] Furthermore, according to embodiments of the present invention, a computer program is also provided, which, when run by a computer or processor, performs the corresponding steps of the web-based continuous frame point cloud data annotation method 100 as described above.

[0083] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0084] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0085] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of the invention. However, this approach should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, its inventive point lies in solving the corresponding technical problem with fewer features than all of those in a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0086] Those skilled in the art will understand that, apart from the mutual exclusion of features, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or apparatus so disclosed can be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0087] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the claims, any of the claimed embodiments can be used in any combination.

[0088] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0089] The above description is merely a specific embodiment of the present invention or an illustration of a specific embodiment. The scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A web-based continuous frame point cloud data annotation method, characterized in that, The method includes: Obtain a labeling request, and in response to the labeling request, download the continuous frame point cloud data to be labeled and the corresponding 2D image data and store them in a local database; Obtain initial bounding boxes for objects in one frame of point cloud data from the continuous frame point cloud data, and automatically calculate final bounding boxes based on the initial bounding boxes; and Map the final bounding box to the 2D image data; The final annotation box is automatically calculated based on the initial annotation box, including: Obtain the point cloud coordinates of the point cloud set within the initial annotation box; and The minimum bounding rectangle of the object is calculated based on the point cloud coordinates, and the initial annotation box is automatically modified according to the minimum bounding rectangle to obtain the final annotation box; The calculation of the minimum bounding rectangle of the object based on the point cloud coordinates includes: The height of the object is estimated based on the point cloud coordinates, and the updated point cloud set within the initial bounding box is calculated based on the estimated height of the object. The updated point cloud set is used to compute an image representation of the object; and The size of the minimum bounding rectangle is calculated based on the image representation of the object, thereby obtaining the minimum bounding rectangle.

2. The method as described in claim 1, characterized in that, The method further includes: downloading continuous frame point cloud data and 2D image data to be labeled and obtaining the initial label box simultaneously using different processing resources.

3. The method as described in claim 1, characterized in that, The local database is the browser's local database.

4. The method as described in claim 1, characterized in that, The method further includes: transforming the point cloud coordinates from the radar coordinate system to the point cloud coordinate system before estimating the height of the object based on the point cloud coordinates.

5. The method as described in claim 4, characterized in that, Estimating the height of the object based on the point cloud coordinates includes: Calculate the height difference between the highest and lowest points in the point cloud set; and The height of the object is estimated by removing the ground height from the height difference, wherein calculating an updated point cloud set within the initial bounding box based on the estimated object height includes: removing points outside the estimated object height range from the point cloud set to obtain the updated point cloud set.

6. The method as described in claim 1, characterized in that, The calculation of the image representation of the object using the updated point cloud set includes: The updated point cloud set is transformed to an image coordinate system, and an image representation of the object is calculated based on the transformed point cloud set.

7. The method as described in claim 1, characterized in that, The method further includes: after labeling the continuous frame point cloud data, performing multi-frame fitting and playback on the labeling results.

8. A web-based continuous frame point cloud data annotation device, characterized in that, The device includes a memory and a processor, the memory storing a computer program executed by the processor, the computer program, when executed by the processor, causing the processor to perform the web-based continuous frame point cloud data annotation method as described in any one of claims 1-7.

9. A computer-readable medium, characterized in that, The computer-readable medium stores computer-executable instructions, which, when executed, perform the web-based continuous frame point cloud data annotation method as described in any one of claims 1-7.

Images