Three-dimensional detection frame processing method, device and equipment based on self-supervised learning

Abstract

The invention provides a three-dimensional detection frame processing method, device and equipment based on self-supervised learning, and the method comprises the steps: obtaining image iteration data and point cloud iteration data, corresponding to an environment scene, of an intelligent vehicle at a current moment in the driving process of the intelligent vehicle; based on the image iteration data, determining first annotation information of a three-dimensional detection frame corresponding to the detection target so as to process and obtain a two-dimensional image contour coordinate of the detection target; based on the point cloud iteration data, determining second labeling information of a three-dimensional detection frame corresponding to the detection target so as to process and obtain a two-dimensional point cloud contour coordinate of the detection target; and performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target labeling information of a three-dimensional detection frame corresponding to the detection target. Therefore, the target labeling information of the three-dimensional detection frame is accurately obtained, manual verification operation is not needed in the whole process, and the labeling efficiency is effectively improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the technical field of self-supervised learning, and in particular, to a method, apparatus and device for processing a three-dimensional detection frame based on self-supervised learning. Background Art

[0002] During vehicle driving, by identifying targets in the driving environment and using 3D detection frames to mark the target information, the specific position of the target can be effectively located, facilitating safe driving of the vehicle.

[0003] In related technologies, most of them use post-fusion or cross-validation to achieve automatic labeling of 3D detection boxes. In different scenarios, post-fusion contains endless scene-specific rules, and cross-validation cannot really make it clear whether to believe in point cloud detection results or image detection results. Even if the model results can be guaranteed to be correct, the alignment problems of different modalities will cause inconsistencies in labeling results, which require manual verification.

[0004] However, the existing method requires manual annotation and proofreading during implementation, which has high annotation costs and a certain error rate in manual annotation, thereby reducing annotation efficiency. Summary of the invention

[0005] The embodiments described herein provide a method, apparatus, and device for processing a three-dimensional detection frame based on self-supervised learning, which overcome the above-mentioned problems.

[0006] In a first aspect, according to the content of the present disclosure, a method for processing a three-dimensional detection frame based on self-supervised learning is provided, comprising:

[0007] During the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment are obtained, wherein the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle;

[0008] Based on the image iteration data, determine first annotation information of a three-dimensional detection frame corresponding to the detection target, so as to process and obtain two-dimensional image contour coordinates of the detection target, wherein the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data;

[0009] Based on the point cloud iteration data, second annotation information of the three-dimensional detection box corresponding to the detection target is determined to obtain the two-dimensional point cloud contour coordinates of the detection target, and the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data;

[0010] Performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of a three-dimensional detection frame corresponding to the detection target;

[0011] The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0012] In a second aspect, according to the content of the present disclosure, a three-dimensional detection frame processing device based on self-supervised learning is provided, comprising:

[0013] An acquisition module is used to acquire, during the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment, wherein the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle;

[0014] A first determination module is used to determine first annotation information of a three-dimensional detection box corresponding to a detection target based on the image iteration data, so as to process and obtain two-dimensional image contour coordinates of the detection target, wherein the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the image iteration data;

[0015] A second determination module is used to determine second annotation information of the three-dimensional detection box corresponding to the detection target based on the point cloud iteration data, so as to process and obtain the two-dimensional point cloud contour coordinates of the detection target, wherein the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data;

[0016] A learning module, used for performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target;

[0017] The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0018] In a third aspect, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of a three-dimensional detection frame processing method based on self-supervised learning in any of the above embodiments are implemented.

[0019] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of a three-dimensional detection frame processing method based on self-supervised learning in any of the above embodiments are implemented.

[0020] The three-dimensional detection frame processing method based on self-supervised learning provided in the embodiment of the present application obtains image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment during the driving process of the intelligent vehicle, the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle; based on the image iteration data, first annotation information of the three-dimensional detection frame corresponding to the detection target is determined to process and obtain the two-dimensional image contour coordinates of the detection target, the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data; based on the point cloud iteration data, second annotation information of the three-dimensional detection frame corresponding to the detection target is determined to process and obtain the two-dimensional point cloud contour coordinates of the detection target, the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the point cloud iteration data; self-supervised learning is performed on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain the target annotation information of the three-dimensional detection frame corresponding to the detection target; wherein the target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame. In this way, by combining multi-view image data and laser point cloud data, a multimodal cross-iterative self-supervision method is adopted to perform self-supervised learning on the two-dimensional boundary contour coordinates in each iteration process, so as to accurately obtain the target annotation information of the three-dimensional detection frame. No manual verification operation is required in the entire process, which effectively improves the annotation efficiency.

[0021] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments will be briefly described below. It should be noted that the drawings described below only relate to some embodiments of the present disclosure, but are not intended to limit the present disclosure, wherein:

[0023] Figure 1 It is a flowchart of a three-dimensional detection frame processing method based on self-supervised learning provided by the present invention.

[0024] Figure 2 It is a structural schematic diagram of a three-dimensional detection frame processing device based on self-supervised learning provided by the present invention.

[0025] Figure 3 It is a structural schematic diagram of a computer device provided by the present disclosure.

[0026] It should be noted that the elements in the drawings are schematic and not drawn to scale. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work also fall within the scope of protection of the present disclosure.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by a person skilled in the art to which the subject matter of the present disclosure belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the specification and the relevant art, and will not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, a statement that two or more parts are "connected" or "coupled" together shall mean that the parts are joined together directly or through one or more intermediate components.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiments" in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists, A and B exist at the same time, and B exists. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. Terms such as "first" and "second" are only used to distinguish one component (or a part of a component) from another component (or another part of a component).

[0031] In the description of the present application, unless otherwise specified, "plurality" means more than two (including two), and similarly, "multiple groups" means more than two groups (including two).

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0033] Figure 1 is a flowchart of a three-dimensional detection frame processing method based on self-supervised learning provided by an embodiment of the present disclosure, such as Figure 1 As shown in FIG. 1 , the specific process of the 3D detection frame processing method based on self-supervised learning includes:

[0034] S110. During the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment are obtained.

[0035] Among them, the image iteration data is used to describe the multi-perspective image data of the environmental scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the same environmental scene collected by the intelligent vehicle at the current moment.

[0036] S120 . Determine first annotation information of a three-dimensional detection frame corresponding to the detection target based on the image iteration data, so as to obtain the two-dimensional image contour coordinates of the detection target.

[0037] The two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data. The two-dimensional image contour coordinates may include: the position coordinates of all two-dimensional boundary contour points of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data.

[0038] It should be noted that the first annotation information is the true value data of the three-dimensional detection frame determined by the image iteration data at the current moment, which includes the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0039] In some embodiments, determining first annotation information of a three-dimensional detection box corresponding to a detection target based on image iteration data to obtain two-dimensional image contour coordinates of the detection target includes:

[0040] Perform sparse visual processing on the image iteration data to obtain a three-dimensional image vector representation corresponding to the image iteration data; decode the three-dimensional image vector representation corresponding to the image iteration data to obtain first annotation information of a three-dimensional detection box corresponding to the detection target; perform boundary distance rendering on the first annotation information of the three-dimensional detection box corresponding to the detection target to obtain two-dimensional image contour coordinates of the detection target.

[0041] Among them, when performing sparse visual processing on image iteration data, the image iteration data can be input into the sparse model, and the initial sparse queries are subjected to cross-attention calculation with multi-view image features (i.e., image iteration data) (image features are used as the key and value of the attention module, sparse queries are used as the query input to the attention module, and attention calculation uses QKV operation), and 3D query vector queries (i.e., the three-dimensional image vector representation corresponding to the image iteration data) are output. The decoder is used to decode the three-dimensional image vector representation corresponding to the image iteration data to obtain 3dbbox (i.e., the first annotation information).

[0042] Thus, by processing the image iteration data, the true value of the 3D detection box at the current moment is obtained, so that the true value of the 3D detection box is rendered into two dimensions to obtain the two-dimensional image contour coordinates.

[0043] In some embodiments, performing boundary distance rendering on the first annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional image contour coordinates of the detection target includes:

[0044] The first annotation information is subjected to boundary distance prediction to obtain an image distance representation sequence between the detection target and the corresponding three-dimensional detection frame; the image distance representation sequence between the detection target and the corresponding three-dimensional detection frame is subjected to two-dimensional boundary rendering to obtain two-dimensional image contour coordinates of the detection target.

[0045] For example, the function value of the first annotation information is calculated using SDF (sign distance function), and the difference between the function value of the 3D real boundary of the detection target and the 3D detection box SDF of the target (obtained by calculating the distance between each gridded point on the 6 boundary surfaces of the 3dbbox and a reference point) is regressed through the residual regression network to obtain the image distance representation sequence. Render it to 2D to obtain the two-dimensional image contour coordinates of the detection target. Thus, the two-dimensional image contour coordinates of the detection target are effectively obtained.

[0046] S130 . Determine second annotation information of a three-dimensional detection box corresponding to the detection target based on the point cloud iteration data, so as to obtain the two-dimensional point cloud contour coordinates of the detection target.

[0047] The two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data. The two-dimensional point cloud contour coordinates may include: the position coordinates of all two-dimensional boundary contour points of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data.

[0048] It should be noted that the second annotation information is the true value data of the three-dimensional detection frame determined by the point cloud iteration data at the current moment, which includes the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0049] In some embodiments, based on the point cloud iteration data, determining the second annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target includes:

[0050] Sparse visual processing is performed on the point cloud iteration data to obtain a three-dimensional image vector representation corresponding to the point cloud iteration data; the three-dimensional image vector representation corresponding to the point cloud iteration data is decoded to obtain second annotation information of a three-dimensional detection box corresponding to the detection target; boundary distance rendering is performed on the second annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target.

[0051] Among them, when performing sparse visual processing on the point cloud iteration data, the point cloud iteration data can be input into the sparse model, and the initial sparse queries are cross-attention calculated with the point cloud features (i.e., the point cloud iteration data) (the point cloud features are used as the key and value of the attention module, the sparse queries are used as the query input to the attention module, and the attention calculation uses QKV operation), and the 3D query vector queries (i.e., the three-dimensional image vector representation corresponding to the point cloud iteration data) are output. The decoder is used to decode the three-dimensional image vector representation corresponding to the point cloud iteration data to obtain 3dbbox (i.e., the second annotation information).

[0052] Thus, by processing the iterative data of the point cloud, the true value of the 3D detection box at the current moment is obtained, and the true value of the 3D detection box is rendered into two dimensions to obtain the two-dimensional point cloud contour coordinates.

[0053] In some embodiments, performing boundary distance rendering on the second annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target includes:

[0054] The boundary distance of the second annotation information is predicted to obtain a point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame; the point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame is rendered in two dimensions to obtain the two-dimensional point cloud contour coordinates of the detection target.

[0055] For example, the function value of the second annotation information is calculated using SDF (sign distance function), and the difference between the function value of the 3D real boundary of the detection target and the 3D detection box SDF of the target (obtained by calculating the distance between each gridded point on the 6 boundary surfaces of the 3D bbox and a reference point) is regressed through the residual regression network to obtain a point cloud distance representation sequence. Render it to 2D to obtain the 2D point cloud contour coordinates of the detection target. Thus, the 2D point cloud contour coordinates of the detection target are effectively obtained.

[0056] S140, performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target;

[0057] The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0058] It should be noted that the target annotation information of the 3D detection frame corresponding to the detection target is the most accurate true value data of the 3D detection frame iterated based on the data at multiple moments. It may be the first annotation information or the second annotation information determined at the current moment as mentioned above, or it may be other annotation information determined at the next moment or subsequent moment.

[0059] In some embodiments, self-supervised learning is performed on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection box corresponding to the detection target, including:

[0060] Perform image instance segmentation on the image iteration data to obtain a two-dimensional instance mask corresponding to the image iteration data, and perform point cloud instance segmentation on the point cloud iteration data to obtain a two-dimensional instance mask corresponding to the point cloud iteration data; use the two-dimensional instance mask corresponding to the image iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, to obtain the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target; use the two-dimensional instance mask corresponding to the point cloud iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, to obtain the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target; fuse the image iteration detection frame true value and the point cloud iteration detection frame true value to obtain the target annotation information of the three-dimensional detection frame corresponding to the detection target.

[0061] Among them, the image instance segmentation model can be used to segment the image iteration data to obtain the two-dimensional instance mask corresponding to the image iteration data, and the point cloud instance segmentation model can be used to segment the point cloud iteration data to obtain the two-dimensional instance mask corresponding to the point cloud iteration data.

[0062] When the two-dimensional image contour coordinates are verified for accuracy using the two-dimensional instance mask corresponding to the image iteration data, if the coordinate difference between the object contour coordinates reflected by the two-dimensional instance mask corresponding to the image iteration data and the two-dimensional image contour coordinates is lower than a preset threshold, then it is determined that the verification result of the two-dimensional image contour coordinates verified using the two-dimensional instance mask corresponding to the image iteration data is a passed verification; if the coordinate difference between the object contour coordinates reflected by the two-dimensional instance mask corresponding to the image iteration data and the two-dimensional image contour coordinates is higher than or equal to the preset threshold, then it is determined that the verification result of the two-dimensional image contour coordinates verified using the two-dimensional instance mask corresponding to the image iteration data is a failed verification.

[0063] When the two-dimensional instance mask corresponding to the point cloud iteration data is used to perform accuracy verification on the two-dimensional point cloud contour coordinates, if the coordinate difference between the object contour coordinates reflected by the two-dimensional instance mask corresponding to the point cloud iteration data and the two-dimensional point cloud contour coordinates is lower than the preset threshold, then it is determined that the two-dimensional instance mask corresponding to the point cloud iteration data is used to perform accuracy verification on the two-dimensional point cloud contour coordinates, and the verification result is that the verification passed; if the coordinate difference between the object contour coordinates reflected by the two-dimensional instance mask corresponding to the point cloud iteration data and the two-dimensional point cloud contour coordinates is higher than or equal to the preset threshold, then it is determined that the two-dimensional instance mask corresponding to the point cloud iteration data is used to perform accuracy verification on the two-dimensional point cloud contour coordinates and the verification result is that the verification failed.

[0064] Fusion of the image iterative detection frame true value and the point cloud iterative detection frame true value to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target may include: using the same / preset different fusion ratios of the image iterative detection frame true value and the point cloud iterative detection frame true value, fusing the image iterative detection frame true value and the point cloud iterative detection frame true value to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target.

[0065] Therefore, by performing multimodal fusion on the obtained image iterative detection box true value and the point cloud iterative detection box true value, the target annotation information of the three-dimensional detection box corresponding to the detection target can be effectively and accurately obtained.

[0066] In some embodiments, the two-dimensional instance mask corresponding to the image iteration data is used to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, to obtain the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target, including:

[0067] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates are verified to be passed based on the two-dimensional instance mask corresponding to the image iteration data, the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is determined based on the first annotation information and the second annotation information. If the first annotation information and the second annotation information are fused at different ratios (the fusion ratio of the first annotation information is higher than the fusion ratio of the second annotation information), the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is obtained.

[0068] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the image iteration data are respectively failed, continue to iterate at the moment until the two-dimensional instance mask corresponding to the image iteration data determined at the current moment, and the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment are passed, then the image iteration detection box true value of the three-dimensional detection box corresponding to the detection target is obtained.

[0069] For example, if the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the image iteration data are respectively failed, then continue to execute to obtain the image iteration data and point cloud iteration data corresponding to the environmental scene of the intelligent vehicle at the current moment; based on the image iteration data, determine the first annotation information of the three-dimensional detection frame corresponding to the detection target to process and obtain the two-dimensional image contour coordinates of the detection target; based on the point cloud iteration data, determine the second annotation information of the three-dimensional detection frame corresponding to the detection target to process and obtain the two-dimensional point cloud contour coordinates of the detection target, until the two-dimensional instance mask corresponding to the image iteration data determined at the current moment, respectively passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then it is determined that the true value of the image iteration detection frame with the most accuracy of the three-dimensional detection frame is obtained.

[0070] In some embodiments, the two-dimensional instance mask corresponding to the point cloud iteration data is used to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, to obtain the point cloud iteration detection box true value of the three-dimensional detection box corresponding to the detection target, including:

[0071] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates are verified to be passed based on the two-dimensional instance mask corresponding to the point cloud iteration data, the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is determined based on the first annotation information and the second annotation information. If the first annotation information and the second annotation information are fused at different ratios (the fusion ratio of the second annotation information is higher than the fusion ratio of the first annotation information), the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is obtained.

[0072] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the point cloud iteration data are respectively failed, continue to iterate at the moment until the two-dimensional instance mask corresponding to the point cloud iteration data determined at the current moment, and the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment are respectively passed, then the point cloud iteration detection box true value of the three-dimensional detection box corresponding to the detection target is obtained.

[0073] For example, if the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the point cloud iteration data are respectively failed, then continue to execute to obtain the image iteration data and point cloud iteration data corresponding to the environmental scene of the intelligent vehicle at the current moment; based on the image iteration data, determine the first annotation information of the three-dimensional detection frame corresponding to the detection target to process and obtain the two-dimensional image contour coordinates of the detection target; based on the point cloud iteration data, determine the second annotation information of the three-dimensional detection frame corresponding to the detection target to process and obtain the two-dimensional point cloud contour coordinates of the detection target, until the two-dimensional instance mask corresponding to the point cloud iteration data determined at the current moment, respectively passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then it is determined that the point cloud iteration detection frame true value with the most accurate three-dimensional detection frame is obtained.

[0074] In this embodiment, during the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment are obtained, the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle; based on the image iteration data, the first annotation information of the three-dimensional detection frame corresponding to the detection target is determined to process and obtain the two-dimensional image contour coordinates of the detection target, and the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data; based on the point cloud iteration data, the second annotation information of the three-dimensional detection frame corresponding to the detection target is determined to process and obtain the two-dimensional point cloud contour coordinates of the detection target, and the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the point cloud iteration data; self-supervised learning is performed on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain the target annotation information of the three-dimensional detection frame corresponding to the detection target; wherein the target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame. In this way, by combining multi-view image data and laser point cloud data, a multimodal cross-iterative self-supervision method is adopted to perform self-supervised learning on the two-dimensional boundary contour coordinates in each iteration process, so as to accurately obtain the target annotation information of the three-dimensional detection frame. No manual verification operation is required in the entire process, which effectively improves the annotation efficiency.

[0075] In summary, this embodiment implements a dual-channel cross-validation 3D detection box automatic labeling paradigm, which can make the detection boxes generated by point cloud and image complement and correct each other through self-supervision. The spatial transformation module is implemented to realize the query alignment of point cloud and image modules, avoiding gradient errors or jitters caused by misalignment of camera and radar. The difference between the 3D detection box and the real 3D boundary of the object is filled by the 3DSDF residual, and it is projected to the image plane to establish consistency constraints with the 2D segmentation mask. The 3D point cloud segmentation results and 2D image segmentation are obtained through the large model, and the results are cross-supervised after projection.

[0076] Figure 2 A structural schematic diagram of a three-dimensional detection frame processing device based on self-supervised learning provided in this embodiment. The three-dimensional detection frame processing device based on self-supervised learning may include: an acquisition module 210, a first determination module 220, a second determination module 230 and a learning module 240.

[0077] The acquisition module 210 is used to obtain the image iteration data and point cloud iteration data corresponding to the environmental scene of the intelligent vehicle at the current moment during the driving process of the intelligent vehicle. The image iteration data is used to describe the multi-perspective image data of the environmental scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environmental scene collected by the intelligent vehicle.

[0078] The first determination module 220 is used to determine the first annotation information of the three-dimensional detection box corresponding to the detection target based on the image iteration data, so as to process and obtain the two-dimensional image contour coordinates of the detection target. The two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the image iteration data.

[0079] The second determination module 230 is used to determine the second annotation information of the three-dimensional detection box corresponding to the detection target based on the point cloud iteration data, so as to process and obtain the two-dimensional point cloud contour coordinates of the detection target. The two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data.

[0080] The learning module 240 is used to perform self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection box corresponding to the detection target.

[0081] The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

[0082] In this embodiment, optionally, the first determining module 220 includes:

[0083] The first processing unit is used to perform sparse visual processing on the image iteration data to obtain a three-dimensional image vector representation corresponding to the image iteration data.

[0084] The first decoding unit is used to decode the three-dimensional image vector representation corresponding to the image iteration data to obtain first annotation information of the three-dimensional detection box corresponding to the detection target.

[0085] The first rendering unit is used to perform boundary distance rendering on the first annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional image contour coordinates of the detection target.

[0086] In this embodiment, optionally, the first rendering unit is specifically configured to:

[0087] The first annotation information is subjected to boundary distance prediction to obtain an image distance representation sequence between the detection target and the corresponding three-dimensional detection frame; the image distance representation sequence between the detection target and the corresponding three-dimensional detection frame is subjected to two-dimensional boundary rendering to obtain two-dimensional image contour coordinates of the detection target.

[0088] In this embodiment, optionally, the second determining module 230 includes:

[0089] The second processing unit is used to perform sparse visual processing on the point cloud iteration data to obtain a three-dimensional image vector representation corresponding to the point cloud iteration data.

[0090] The second decoding unit is used to decode the three-dimensional image vector representation corresponding to the point cloud iteration data to obtain second annotation information of the three-dimensional detection box corresponding to the detection target.

[0091] The second rendering unit is used to perform boundary distance rendering on the second annotation information of the three-dimensional detection box corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target.

[0092] In this embodiment, optionally, the second rendering unit is specifically configured to:

[0093] The boundary distance of the second annotation information is predicted to obtain a point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame; the point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame is rendered in two dimensions to obtain the two-dimensional point cloud contour coordinates of the detection target.

[0094] In this embodiment, optionally, the learning module 240 includes:

[0095] The segmentation unit is used to perform image instance segmentation on the image iteration data to obtain a two-dimensional instance mask corresponding to the image iteration data, and to perform point cloud instance segmentation on the point cloud iteration data to obtain a two-dimensional instance mask corresponding to the point cloud iteration data.

[0096] The verification unit is used to use the two-dimensional instance mask corresponding to the image iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, so as to obtain the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target; and use the two-dimensional instance mask corresponding to the point cloud iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates, respectively, so as to obtain the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target.

[0097] The fusion unit is used to fuse the true value of the image iterative detection box and the true value of the point cloud iterative detection box to obtain the target annotation information of the three-dimensional detection box corresponding to the detection target.

[0098] In this embodiment, the verification unit is optionally used to:

[0099] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the image iteration data are passed, then the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is determined based on the first annotation information and the second annotation information; if the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the image iteration data are failed, the moment iteration is continued until the two-dimensional instance mask corresponding to the image iteration data determined at the current moment passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is obtained.

[0100] In this embodiment, the verification unit is optionally used to:

[0101] If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the point cloud iteration data are passed, then based on the first annotation information and the second annotation information, the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is determined; if the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the point cloud iteration data are failed, the moment iteration is continued until the two-dimensional instance mask corresponding to the point cloud iteration data determined at the current moment passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is obtained.

[0102] The three-dimensional detection frame processing device based on self-supervised learning provided in the present disclosure can execute the above method embodiments. Its specific implementation principles and technical effects can be found in the above method embodiments, and the present disclosure will not repeat them here.

[0103] The present application also provides a computer device. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0104] The computer device includes a memory 310 and a processor 320 that are connected to each other through a system bus. It should be noted that the figure only shows a computer device with a memory 310 and a processor 320, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0105] Computer devices can be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. Computer devices can interact with users through keyboards, mice, remote controls, touch pads, or voice control devices.

[0106] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes a non-volatile memory or a volatile memory, for example, a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc., and the RAM may include a static RAM or a dynamic RAM. In some embodiments, the memory 310 may be an internal storage unit of a computer device, for example, a hard disk or a memory of the computer device. In other embodiments, the memory 310 may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, or a flash card (FlashCard) equipped on the computer device. Of course, the memory 310 may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device, such as the program code of the above method. In addition, the memory 310 may also be used to temporarily store various data that have been output or are to be output.

[0107] The processor 320 is generally used to perform the overall operation of the computer device. In this embodiment, the memory 310 is used to store program codes or instructions, the program code includes computer operation instructions, and the processor 320 is used to execute the program codes or instructions stored in the memory 310 or process data, such as running the program code of the above method.

[0108] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus system can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0109] Another embodiment of the present application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads a computer-readable program code stored in the computer-readable medium, so that the processor can execute the functional actions specified in each step or a combination of steps in the above method; and generate a device for implementing the functional actions specified in each block or a combination of blocks in the block diagram.

[0110] Computer-readable media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any appropriate combination of the foregoing, the memory is used to store program codes or instructions, the program codes include computer operating instructions, and the processor is used to execute the program codes or instructions of the above methods stored in the memory.

[0111] For the definitions of memory and processor, please refer to the description of the aforementioned computer device embodiment and will not be repeated here.

[0112] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0113] Each functional unit or module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0114] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program code.

[0115] In the claims, any reference symbols placed between brackets shall not be construed as limiting the claims. The word "comprising" described in the present application 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 present application can be implemented with the aid of hardware comprising several different elements and with the aid of a suitably programmed computer. In a unit claim that lists a number of devices, several units of these devices may be embodied by the same hardware item. The use of first, second, and third, etc. does not indicate any order, and these words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

[0116] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A three-dimensional detection frame processing method based on self-supervised learning, characterized in that: include: During the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment are obtained, wherein the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle; Based on the image iteration data, determine first annotation information of a three-dimensional detection frame corresponding to the detection target, so as to process and obtain two-dimensional image contour coordinates of the detection target, wherein the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection frame corresponding to the detection target determined by the image iteration data; Based on the point cloud iteration data, second annotation information of the three-dimensional detection box corresponding to the detection target is determined to obtain the two-dimensional point cloud contour coordinates of the detection target, and the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data; Performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of a three-dimensional detection frame corresponding to the detection target; The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

2. The method according to claim 1, characterized in that The determining, based on the image iteration data, first annotation information of a three-dimensional detection frame corresponding to the detection target to obtain the two-dimensional image contour coordinates of the detection target includes: Performing sparse visual processing on the image iteration data to obtain a three-dimensional image vector representation corresponding to the image iteration data; Decoding the three-dimensional image vector representation corresponding to the image iteration data to obtain first annotation information of the three-dimensional detection box corresponding to the detection target; Boundary distance rendering is performed on the first annotation information of the three-dimensional detection frame corresponding to the detection target to obtain the two-dimensional image contour coordinates of the detection target.

3. The method according to claim 2, characterized in that The performing boundary distance rendering on the first annotation information of the three-dimensional detection frame corresponding to the detection target to obtain the two-dimensional image contour coordinates of the detection target includes: Performing boundary distance prediction on the first annotation information to obtain an image distance representation sequence between the detection target and the corresponding three-dimensional detection frame; Perform two-dimensional boundary rendering on the image distance representation sequence between the detection target and the corresponding three-dimensional detection frame to obtain the two-dimensional image contour coordinates of the detection target.

4. The method according to claim 1, characterized in that: The step of determining second annotation information of a three-dimensional detection frame corresponding to the detection target based on the point cloud iteration data to obtain the two-dimensional point cloud contour coordinates of the detection target includes: Performing sparse visual processing on the point cloud iteration data to obtain a three-dimensional image vector representation corresponding to the point cloud iteration data; Decoding the three-dimensional image vector representation corresponding to the point cloud iteration data to obtain second annotation information of the three-dimensional detection box corresponding to the detection target; Boundary distance rendering is performed on the second annotation information of the three-dimensional detection frame corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target.

5. The method according to claim 4, characterized in that The performing boundary distance rendering on the second annotation information of the three-dimensional detection frame corresponding to the detection target to obtain the two-dimensional point cloud contour coordinates of the detection target includes: Performing boundary distance prediction on the second annotation information to obtain a point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame; Perform two-dimensional boundary rendering on a point cloud distance representation sequence between the detection target and the corresponding three-dimensional detection frame to obtain two-dimensional point cloud contour coordinates of the detection target.

6. The method according to claim 1, characterized in that The self-supervised learning of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target includes: Performing image instance segmentation on the image iteration data to obtain a two-dimensional instance mask corresponding to the image iteration data, and performing point cloud instance segmentation on the point cloud iteration data to obtain a two-dimensional instance mask corresponding to the point cloud iteration data; Using the two-dimensional instance mask corresponding to the image iteration data, respectively, the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates are accuracy checked to obtain the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target; Using the two-dimensional instance mask corresponding to the point cloud iteration data, respectively perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target; The true value of the iterative detection box of the image and the true value of the iterative detection box of the point cloud are fused to obtain target annotation information of the three-dimensional detection box corresponding to the detection target.

7. The method according to claim 6, characterized in that The method of using the two-dimensional instance mask corresponding to the image iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates respectively to obtain the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target includes: If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates are respectively verified to be passed based on the two-dimensional instance mask corresponding to the image iteration data, then determining the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target based on the first annotation information and the second annotation information; If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the image iteration data are respectively failed, continue to iterate at the moment until the two-dimensional instance mask corresponding to the image iteration data determined at the current moment, respectively, passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then the image iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target is obtained.

8. The method according to claim 6, characterized in that The method of using the two-dimensional instance mask corresponding to the point cloud iteration data to perform accuracy verification on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates respectively to obtain the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target includes: If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates are respectively verified to be passed based on the two-dimensional instance mask corresponding to the point cloud iteration data, then determining the point cloud iteration detection frame true value of the three-dimensional detection frame corresponding to the detection target based on the first annotation information and the second annotation information; If the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates based on the two-dimensional instance mask corresponding to the point cloud iteration data are respectively failed, continue to iterate at the moment until the two-dimensional instance mask corresponding to the point cloud iteration data determined at the current moment, respectively passes the accuracy verification results of the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates at the current moment, then the point cloud iteration detection box true value of the three-dimensional detection box corresponding to the detection target is obtained.

9. A three-dimensional detection frame processing device based on self-supervised learning, characterized in that: include: An acquisition module is used to acquire, during the driving process of the intelligent vehicle, image iteration data and point cloud iteration data corresponding to the environment scene of the intelligent vehicle at the current moment, wherein the image iteration data is used to describe the multi-view image data of the environment scene collected by the intelligent vehicle at the current moment, and the point cloud iteration data is used to describe the laser point cloud data of the environment scene collected by the intelligent vehicle; A first determination module is used to determine first annotation information of a three-dimensional detection box corresponding to a detection target based on the image iteration data, so as to process and obtain two-dimensional image contour coordinates of the detection target, wherein the two-dimensional image contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the image iteration data; A second determination module is used to determine second annotation information of the three-dimensional detection box corresponding to the detection target based on the point cloud iteration data, so as to process and obtain the two-dimensional point cloud contour coordinates of the detection target, wherein the two-dimensional point cloud contour coordinates are used to describe the two-dimensional boundary contour coordinates of the three-dimensional detection box corresponding to the detection target determined by the point cloud iteration data; A learning module, used for performing self-supervised learning on the two-dimensional image contour coordinates and the two-dimensional point cloud contour coordinates to obtain target annotation information of the three-dimensional detection frame corresponding to the detection target; The target annotation information of the three-dimensional detection frame corresponding to the detection target includes: the center point coordinates, length, width, height and orientation of the three-dimensional detection frame.

10. A computer device, characterized in that: The invention comprises a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, a three-dimensional detection frame processing method based on self-supervised learning as described in any one of claims 1 to 8 is implemented.

Images