CN116052097A

CN116052097A - Map element detection method and device, electronic equipment and storage medium

Info

Publication number: CN116052097A
Application number: CN202211584273.6A
Authority: CN
Inventors: 张瀚天; 周尧; 万国伟; 张晔
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2022-12-09
Filing date: 2022-12-09
Publication date: 2023-05-02

Abstract

The disclosure provides a map element detection method, a map element detection device, electronic equipment and a storage medium, relates to the technical field of artificial intelligence, and particularly relates to an automatic driving technology and a high-precision map technology. The specific implementation scheme comprises the following steps: acquiring a multi-view image including surrounding information of an automatically driven vehicle; extracting image features from the multi-view image, and converting the image features into bird's-eye view features under the bird's-eye view angle; based on the aerial view feature, map elements around the autonomous vehicle are determined by means of example segmentation. The method and the system can accurately obtain the information of the map element example level, and can effectively assist the planning decision of the automatic driving vehicle.

Description

Map element detection method and device, electronic equipment and storage medium

Technical Field

The present disclosure relates to the field of artificial intelligence technology, and in particular, to an automatic driving technology and a high-precision map technology, and more particularly, to a map element detection method, apparatus, electronic device, storage medium, and computer program product.

Background

With the development of artificial intelligence, automatic driving technology is also rapidly developing. The automatic driving technique refers to a technique in which a vehicle collects information around the vehicle using a sensor and controls the vehicle according to the collected information.

Disclosure of Invention

The present disclosure provides a map element detection method, apparatus, electronic device, storage medium, and computer program product.

According to an aspect of the present disclosure, there is provided a map element detection method including:

acquiring a multi-view image including surrounding information of an automatically driven vehicle;

extracting image features from the multi-view image, and converting the image features into bird's-eye view features under the bird's-eye view;

and determining map elements around the automatic driving vehicle by adopting an example segmentation mode based on the aerial view characteristics.

According to an aspect of the present disclosure, there is provided a map element detection apparatus including:

the data acquisition module is used for acquiring a multi-view image comprising surrounding environment information of the automatic driving vehicle;

the feature extraction module is used for extracting image features from the multi-view image and converting the image features into bird's-eye view features under the bird's-eye view angle;

and the detection module is used for determining map elements around the automatic driving vehicle by adopting an example segmentation mode based on the aerial view characteristics.

According to another aspect of the present disclosure, there is provided an electronic device including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein, the liquid crystal display device comprises a liquid crystal display device,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the map element detection method of any embodiment of the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the map element detection method of any embodiment of the present disclosure.

According to another aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the map element detection method of any embodiment of the present disclosure.

According to the technology disclosed by the invention, the information of the map element instance level can be accurately obtained, and the planning decision of the automatic driving vehicle can be effectively assisted.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.

Drawings

The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

fig. 1 is a flow chart of a map element detection method according to an embodiment of the disclosure;

FIG. 2 is a flow chart of yet another map element detection method provided by an embodiment of the present disclosure;

FIG. 3 is a flow chart of yet another map element detection method provided by an embodiment of the present disclosure;

FIG. 4 is a flow chart of yet another map element detection method provided by an embodiment of the present disclosure;

fig. 5 is a schematic structural diagram of a map element detection apparatus provided in an embodiment of the present disclosure;

fig. 6 is a block diagram of an electronic device for implementing a map element detection method of an embodiment of the present disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made

Modifications may be made without departing from the scope and spirit of the disclosure. Also, descriptions of well-known functions and constructions are omitted in the following 5 description for clarity and conciseness.

Fig. 1 is a flowchart of a map element detection method according to an embodiment of the present disclosure, which is applicable to a situation of identifying map elements around a vehicle in a dynamic driving scenario. The method may be performed by a map element detection apparatus implemented in software and/or hardware and integrated on an electronic device.

0 specifically, referring to fig. 1, the map element detection method includes the following steps:

s101, acquiring a multi-view image comprising surrounding environment information of the automatic driving vehicle.

In this embodiment, in order to understand the driving environment around the vehicle, the autonomous vehicle may configure multiple sensors to collect surrounding environment information, where the multiple sensors include a camera sensor and a laser radar

A sensor, etc. The multi-angle image may be image data photographed from a plurality of angles by a camera sensor, 5, for example, a plurality of camera sensors are arranged around the vehicle, the plurality of camera sensors may photograph

Image data corresponding to a 360 degree scene around a vehicle. On the basis, the multi-view image including the surrounding environment information of the automatic driving vehicle can be acquired through the data input interface. The acquired multi-view images are optionally acquired by the camera sensor under the same time stamp.

S102, extracting image features from the multi-view image, and converting the image features into bird 'S-eye view features under the bird' S-eye view angle 0.

In this embodiment, in the autopilot field, if the perceived information in the ground plane space can be obtained, the dimension conversion is not required to be performed, and the downstream planning decision can be made. Since the multi-view image is two-dimensional data, the multi-view image is optionally converted for use in downstream planning decisions

An image at a bird's eye view angle. In implementation, feature extraction may be performed on the multi-view image first, for example, feature extraction may be performed by using 5 a neural network (for example, a convolutional neural network) capable of extracting features, so as to obtain corresponding image features. The image features may then be converted to Bird's-eye-view features at a Bird's-eye-view angle by visual recognition techniques, such as transform techniques. It should be noted that other techniques may be used to implement the conversion, and the present invention is not limited thereto.

S103, determining map elements around the automatic driving vehicle by adopting an example segmentation mode based on the aerial view features.

On the basis of obtaining the aerial view feature through the step of S102, the aerial view feature may be segmented and identified by using an example segmentation technique to obtain example-level map elements, where the map elements may be selected from lane lines, road edges, crosswalk, intersections, and the like around the autonomous vehicle. It should be noted that, the map elements are not identified by semantic segmentation, because semantic segmentation identification is adopted, only semantic information contained in each pixel can be determined, for example, only the element type of each pixel can be known, and it cannot be determined which specific map elements are composed of which specific pixels, so that the subsequent automatic driving planning decision cannot directly use the identification result.

In this embodiment, the example segmentation is performed on the aerial view feature, so that the example-level map elements around the vehicle can be accurately identified, and then the planning decision of the automatic driving vehicle can be assisted according to the identified map elements.

Fig. 2 is a flow diagram of yet another map element detection method according to an embodiment of the present disclosure. Referring to fig. 2, the map element detection method is as follows:

s201, acquiring a multi-view image comprising surrounding environment information of the automatic driving vehicle.

In this embodiment, in order to accurately and rapidly identify map elements around an automatic self-driving vehicle, a map element detection model is trained in advance; the map element detection model comprises a backbone network (such as a convolutional neural network) for extracting image features of the multi-view image, a feature conversion network (such as a transform network) for converting the image features into bird's eye view features, an instance segmentation network and a multi-layer neural network for predicting segmentation results. Based on the map element detection model, the map element detection is realized according to the steps of S202-S207.

S202, extracting image features from the multi-view image.

Optionally, the multi-view image is used as an input of a backbone network in the map element detection module, and corresponding image characteristics are determined according to the output of the backbone network.

S203, dividing a preset range area around the automatic driving vehicle into a plurality of aerial view grids under the aerial view angle, wherein each aerial view grid is represented by a query parameter query.

In this embodiment, the preset range area around the autonomous vehicle may be set according to actual needs, for example, an area of 50m×50m around the vehicle is used as the preset range area. Furthermore, the preset range area can be divided into a plurality of aerial view grids according to the spatial resolution under the preset aerial view angle, and each aerial view grid is represented by one query parameter query, that is, each query parameter query is only responsible for representing a corresponding small range area.

S204, determining the aerial view features corresponding to each query parameter query through a feature conversion network in the pre-trained map element detection model based on the query parameter query and the image features.

In this embodiment, the feature conversion network in the map element detection model is optionally a transducer network. Each feature of the multi-view image may be represented by a key and a value; where a key may represent the location of an image feature and value represents a specific value of the image feature at that location. When the aerial view features are extracted, the query parameter query and each image feature point key and value are used as input of a transducer network. For any query parameter query, the transformer network can obtain the aerial view feature corresponding to the query parameter query by calculating the similarity or the correlation between the query parameter query and each image feature key and taking the value of the similarity or the correlation as the value corresponding to the key, and further carrying out weighted calculation on the value. And integrating the aerial view features corresponding to all query parameters query to obtain the complete aerial view features corresponding to the multi-view image. It should be noted that, through the transformation network, the transformation of the aerial view feature can be rapidly and accurately improved. On this basis, the map elements may be detected in accordance with the steps of S205 to S207.

S205, taking the aerial view features corresponding to each query parameter query as to-be-detected examples, and carrying out feature aggregation on each to-be-detected example through an attention mechanism in a map element detection model.

In this embodiment, the example segmentation network (e.g., mask2former network) in the map element detection model is also optionally a transformation network with attention mechanisms, such as self-attention and cross-attention mechanisms. Since the bird's eye view feature corresponding to each query parameter query is input into the partition network as an instance to be detected, feature aggregation is performed on each instance to be detected through the self-attention mechanism self-attention and the cross-attention mechanism cross-attention, for example, part of features are extracted from one instance to be detected and are supplemented to another instance to be detected. Thus, the feature aggregation can ensure that the to-be-detected instance can better express a map element.

S206, performing dot product processing on the to-be-detected instance subjected to feature aggregation and the target aerial view feature to obtain a segmentation mask.

In this embodiment, the target aerial view feature is obtained by performing up-sampling determination on the aerial view feature in advance, for example, performing up-sampling on the complete aerial view feature described above. On the basis, the segmentation network only needs to perform dot product operation processing on the to-be-detected examples subjected to feature aggregation and the target aerial view features to obtain a segmentation mask corresponding to each to-be-detected example, wherein the segmentation mask can be a map element or other environment elements, and therefore classification and identification are needed according to the step S207.

S207, determining the type of the map element to which the segmentation mask belongs through a multi-layer neural network in the map element detection model.

In this embodiment, the segmentation mask obtained by segmentation is input into the pre-trained multi-layer neural network, and the map element corresponding to the segmentation mask is determined according to the network output.

In the embodiment, the segmentation and recognition of the map elements are accurately realized through the segmentation network and the multi-layer neural network in the map element detection model, so that the recognized map elements can directly participate in the planning decision of automatic driving.

Fig. 3 is a flow diagram of yet another map element detection method according to an embodiment of the present disclosure. Referring to fig. 3, the map element detection method is as follows:

s301, acquiring a multi-view image comprising surrounding environment information of the automatic driving vehicle.

S302, extracting image features from the multi-view images, and converting the image features into bird 'S-eye view features under the bird' S-eye view angles.

S303, determining map elements around the automatic driving vehicle by adopting an example segmentation mode based on the aerial view features.

Wherein, see the above description for details of the steps S301-S303.

S304, an online high-precision map is constructed according to the identified map elements, and the original high-precision map is updated through the online high-precision map.

On the basis of identifying the map elements, determining the relationship (such as the predecessor relationship of the lane lines) of the map elements according to query parameters query corresponding to the map elements, so that the on-line high-precision map can be built according to the relationship between the identified map elements and the map elements; and the original high-precision map is updated according to the online high-precision map, so that the updating period of the high-precision map is shortened compared with the mode of updating the high-precision map by collecting data through a collection vehicle.

Fig. 4 is a flow diagram of yet another map element detection method according to an embodiment of the present disclosure. Referring to fig. 4, the map element detection method logic is specifically as follows:

s401, constructing tag data according to positioning data and high-precision map data of the automatic driving vehicle, and taking the tag data as a training sample of a training map element detection model.

In this embodiment, to train the map element detection model, a training sample needs to be prepared in advance. Currently, the conventional method determines a training sample by a manual labeling method, so that the sample generation efficiency is extremely low. Based on this, in consideration of the fact that a map element in a high-precision map is labeled in addition to a map image, a method has been proposed in which tag data can be constructed from the high-precision map and positioning data, and the tag data can be used as a training sample. When the method is realized, the positioning data of the automatic driving vehicle can be acquired firstly, and then the target position is determined according to the positioning data of the automatic driving vehicle; acquiring a high-precision map corresponding to the target position, for example, intercepting the high-precision map near the target position; and converting the high-precision map into a camera coordinate system (such as a coordinate system of a camera sensor of an automatic driving vehicle), and converting map elements in the high-precision map into element data of a mask type to obtain tag data, wherein the tag data comprises the high-precision map and the mask data, and the element type corresponds to the mask data.

S402, training a map element detection model based on the training sample.

On the basis of obtaining a training sample, the training process of the map element detection model is as follows: the label data is used as a training sample and is input into an instance segmentation model to obtain an instance mask and a type output by the model; performing Hungary matching on the segmentation mask and the type output by the model and the label; in the Hungary matching, the cost function is calculated mainly according to the class of the output result and the segmentation mask so as to adjust the network parameters of the model according to the loss; the trained loss function loss is divided into two parts, the first part is the loss function for estimating the category (lane line, humanized cross road, road edge, etc.) to which the element belongs, and the second part is the loss function for estimating the segmentation mask, such as cross entropy classification loss.

After training the map element detection model, the map element detection may be performed according to the steps of S402 to S409.

S403, acquiring a multi-view image comprising surrounding environment information of the automatic driving vehicle.

S404, extracting image features from the multi-view image, and converting the image features into bird 'S-eye view features under the bird' S-eye view.

S405, determining map elements around the automatic driving vehicle by adopting an example segmentation mode based on the aerial view features.

S406, dividing a preset range area around the automatic driving vehicle into a plurality of aerial view grids under the aerial view angle, wherein each aerial view grid is represented by a query parameter query.

S407, determining the aerial view features corresponding to each query parameter query through a feature conversion network in the pre-trained map element detection model based on the query parameter query and the image features.

S408, taking the aerial view features corresponding to each query parameter query as to-be-detected examples, and carrying out feature aggregation on each to-be-detected example through an attention mechanism in a map element detection model.

S409, performing dot product processing on the to-be-detected instance subjected to feature aggregation and the target aerial view feature, and obtaining a segmentation mask 5.

The target aerial view feature is determined by up-sampling the aerial view feature in advance.

S410, determining the type of the map element to which the segmentation mask belongs through a multi-layer neural network in the map element detection model.

In this embodiment, the training sample of the model 0 is constructed according to the positioning data and the high-precision map data of the automatic driving vehicle, so that the efficiency of constructing the sample is improved compared with the manual labeling.

Fig. 5 is a schematic structural diagram of a map element detection apparatus according to an embodiment of the present disclosure. The present embodiment is applicable to a case where map elements around a vehicle are identified in a dynamic driving scene. Referring to fig. 5, the apparatus includes:

a data acquisition module 501 for acquiring a multi-view 5-angle image including surrounding information of an automated driving vehicle;

the feature extraction module 502 is configured to extract image features from the multi-view image, and convert the image features into aerial view features under aerial view;

the detection module 503 is configured to determine map elements around the autonomous vehicle by using an example segmentation method based on the aerial view feature.

0 on the basis of the above embodiment, optionally, the feature extraction module is further configured to:

dividing a preset range area around the automatic driving vehicle into a plurality of aerial view grids under the aerial view angle, wherein each aerial view grid is characterized by a query parameter query;

based on the query parameters query and the image features, determining the aerial view features corresponding to each query parameter query through a feature conversion network in the pre-trained map element detection model.

5 on the basis of the above embodiment, optionally, the detection module is further configured to:

taking the aerial view characteristics corresponding to each query parameter query as an instance to be detected, and carrying out characteristic aggregation on each instance to be detected through an attention mechanism in a map element detection model;

performing dot product processing on the to-be-detected instance subjected to feature aggregation and the target aerial view feature to obtain a segmentation mask; the target aerial view features are obtained by carrying out up-sampling determination on the aerial view features in advance;

and determining the type of the map element to which the segmentation mask belongs through a multi-layer neural network in the map element detection model.

On the basis of the above embodiment, optionally, the method further includes:

and the map updating module is used for constructing an online high-precision map according to the identified map elements and updating the original high-precision map through the online high-precision map.

On the basis of the above embodiment, optionally, the method further includes:

and the labeling module is used for constructing label data according to the positioning data and the high-precision map data of the automatic driving vehicle, and taking the label data as a training sample of the training map element detection model.

On the basis of the above embodiment, optionally, the labeling module is further configured to:

determining a target position according to positioning data of the automatic driving vehicle;

acquiring a high-precision map corresponding to the target position;

and converting the high-precision map into a camera coordinate system, and converting map elements in the high-precision map into mask type data to obtain tag data.

The map element detection device provided by the embodiment of the disclosure can execute the map element detection method provided by any embodiment of the disclosure, and has the corresponding functional modules and beneficial effects of the execution method. Reference is made to the description of any method embodiment of the disclosure for details not explicitly described in this embodiment.

In the technical scheme of the disclosure, the acquisition, storage, application and the like of the related user personal information all conform to the regulations of related laws and regulations, and the public sequence is not violated.

According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 6, the apparatus 600 includes a computing unit 601 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM603, various programs and data required for the operation of the device 600 may also be stored. The computing unit 601, ROM602, and RAM603 are connected to each other by a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

Various components in the device 600 are connected to the I/O interface 605, including: an input unit 606 such as a keyboard, mouse, etc.; an output unit 607 such as various types of displays, speakers, and the like; a storage unit 608, such as a magnetic disk, optical disk, or the like; and a communication unit 609 such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.

The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the respective methods and processes described above, such as a map element detection method. For example, in some embodiments, the map element detection method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM602 and/or the communication unit 609. When the computer program is loaded into the RAM603 and executed by the computing unit 601, one or more steps of the map element detection method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the map element detection method by any other suitable means (e.g., by means of firmware).

Various implementations of the systems and techniques described here above can be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.

The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.

Claims

1. A map element detection method, comprising:

2. The method of claim 1, wherein converting the image features to bird's-eye view features at a bird's-eye view angle comprises:

dividing a preset range area around the automatic driving vehicle into a plurality of aerial view grids under an aerial view angle, wherein each aerial view grid is characterized by a query parameter query;

and determining the aerial view features corresponding to each query parameter query through a feature conversion network in a pre-trained map element detection model based on the query parameter query and the image features.

3. The method of claim 2, wherein determining map elements of the autonomous vehicle perimeter using instance segmentation based on the bird's eye view features comprises:

taking the aerial view characteristics corresponding to each query parameter query as to-be-detected examples, and carrying out characteristic aggregation on each to-be-detected example through an attention mechanism in the map element detection model;

performing dot product processing on the to-be-detected instance subjected to feature aggregation and the target aerial view feature to obtain a segmentation mask; the target aerial view feature is obtained by carrying out up-sampling determination on the aerial view feature in advance;

and determining the type of the map element to which the segmentation mask belongs through the multi-layer neural network in the map element detection model.

4. The method of claim 1, further comprising:

and constructing an online high-precision map according to the identified map elements, and updating the original high-precision map through the online high-precision map.

5. The method of claim 2, further comprising:

and constructing tag data according to the positioning data and the high-precision map data of the automatic driving vehicle, and taking the tag data as a training sample for training the map element detection model.

6. The method of claim 5, wherein constructing tag data from the location data and high-precision map data of the autonomous vehicle comprises:

determining a target position according to the positioning data of the automatic driving vehicle;

acquiring a high-precision map corresponding to the target position;

and converting the high-precision map into a camera coordinate system, and converting map elements in the high-precision map into element data of a mask type to obtain the tag data.

7. A map element detection apparatus comprising:

8. The apparatus of claim 7, wherein the feature extraction module is further to:

9. The apparatus of claim 8, wherein the detection module is further to:

10. The apparatus of claim 7, further comprising:

11. The apparatus of claim 8, further comprising:

and the labeling module is used for constructing label data according to the positioning data and the high-precision map data of the automatic driving vehicle, and taking the label data as a training sample for training the map element detection model.

12. The apparatus of claim 11, wherein the labeling module is further to:

acquiring a high-precision map corresponding to the target position;

and converting the high-precision map into a camera coordinate system, and converting map elements in the high-precision map into mask type data to obtain the tag data.

13. An electronic device, comprising:

at least one processor; and

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the map element detection method of any one of claims 1-6.

14. A non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the map element detection method according to any one of claims 1 to 6.

15. A computer program product comprising a computer program which, when executed by a processor, implements the map element detection method according to any one of claims 1-6.