Obstacle attribute discrimination method and device, computer equipment and medium

Abstract

The invention relates to the technical field of automatic driving, in particular to an obstacle attribute judgment method and device, computer equipment and a medium. Mapping the first position information to a vehicle surrounding environment image collected by a camera to obtain second position information of the to-be-processed obstacle in the vehicle surrounding environment image; then determining a target image corresponding to the obstacle from the vehicle surrounding environment image by using the second position information; and finally, inputting the target image into the pedestrian false alarm model to judge the effectiveness of the attribute of the obstacle. According to the invention, secondary attribute discrimination is carried out on the obstacle with the pedestrian attribute within the preset range of the driving vehicle and with the confidence lower than the preset threshold, i.e., the obstacle to be processed, and the features of the target image are read and analyzed more finely by using the pedestrian false alarm model, so that a more accurate discrimination result is obtained. And the stability and the safety in the driving process are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of autonomous driving technology, and in particular to a method, device, computer equipment, and medium for determining obstacle attributes. Background Art

[0002] Autonomous driving, a technology that uses computer systems to achieve unmanned driving, offers the potential to reduce costs, alleviate traffic congestion, increase road capacity, and enhance human productivity. However, it's not currently in practical use due to concerns about driving safety, and the accuracy of obstacle recognition is undoubtedly a crucial factor affecting driving safety.

[0003] Currently, the bird's-eye view prediction models used in autonomous driving sometimes misidentify obstacles as pedestrians, such as weeds or trees growing in roadside grass. Misidentification of obstacles near the vehicle can easily trigger emergency braking, compromising driving safety. Currently, the primary approach to addressing misidentification is through confidence levels. For example, if an obstacle near the vehicle is identified as a person, but the confidence level falls below a preset threshold, the system assumes the obstacle is not a person and ignores it, allowing the vehicle to continue driving. However, using confidence levels alone as the sole basis for misidentification can lead to inaccurate results.

[0004] Therefore, how to accurately determine whether an obstacle is mistakenly identified as a pedestrian becomes an urgent problem that needs to be solved. Summary of the Invention

[0005] In order to solve the above technical problems, the present disclosure provides a method, apparatus, computer equipment and medium for determining obstacle attributes.

[0006] In a first aspect, the present invention provides a method for determining obstacle attributes, comprising:

[0007] Obtain first position information of an obstacle to be processed, where the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, has an obstacle attribute of a pedestrian, and has an attribute confidence level below a preset threshold. Based on internal and external parameters of a vehicle-mounted camera, map the first position information to an image of the vehicle surrounding environment captured by the camera to obtain second position information of the obstacle to be processed in the image of the vehicle surrounding environment. Based on the second position information, determine a target image corresponding to the obstacle to be processed from the image of the vehicle surrounding environment. Input the target image into a pre-built pedestrian false alarm model, which outputs a judgment result. When the judgment result corresponds to the obstacle attribute, the obstacle attribute is determined to be valid.

[0008] The obstacle attribute determination method provided in this embodiment, after obtaining first position information of an obstacle to be processed, first maps the first position information to an image of the vehicle surrounding environment captured by the camera using the internal and external parameters of the vehicle-mounted camera, thereby obtaining second position information of the obstacle to be processed in the image of the vehicle surrounding environment. Then, using the second position information, a target image corresponding to the obstacle is determined from the image of the vehicle surrounding environment. Finally, the target image is input into a pedestrian false alarm model to determine the validity of the obstacle attributes. This embodiment performs a secondary attribute determination on obstacles with pedestrian attributes within a preset range of the driving vehicle and with a confidence level below a preset threshold, namely, obstacles to be processed. The pedestrian false alarm model is used to perform more detailed reading and analysis of the features of the target image, thereby obtaining a more accurate determination result, thereby resolving the defect of misidentification and improving the stability and safety of the driving process.

[0009] In an optional embodiment, obtaining first position information of an obstacle to be processed includes:

[0010] Obtain description information of the target obstacle, including its location information, obstacle attributes, and attribute confidence. The target obstacle is obtained by post-processing the obstacle using a non-maximum suppression algorithm. Target obstacles within a preset range centered on the driving vehicle, whose obstacle attributes are pedestrians and whose attribute confidence is less than a preset threshold, are determined as obstacles to be processed. Location information is extracted from the description information of the obstacle to be processed, and the location information is used as first location information.

[0011] In an optional embodiment, obtaining description information of the target obstacle includes:

[0012] Obtain an image of the vehicle's surroundings; input the image into a pre-built bird's-eye view prediction model to output description information of obstacles; based on the description information of the obstacles, use a non-maximum suppression algorithm to post-process the obstacles to determine the target obstacle from the obstacles and obtain the description information of the target obstacle.

[0013] The obstacle attribute determination method provided in this embodiment uses the NMS algorithm to post-process obstacles, remove some redundant obstacles, and screen out target obstacles with the best bounding box, thereby reducing the amount of obstacle data to a certain extent and improving work efficiency.

[0014] In an optional embodiment, based on internal and external parameters of a vehicle-mounted camera, the first position information is mapped to an image of the vehicle surrounding environment captured by the camera to obtain second position information of the obstacle to be processed in the image of the vehicle surrounding environment, including:

[0015] The first position information is converted into the third position information of the obstacle to be processed in the camera coordinate system using the external parameters of the camera; and the third position information is converted into the second position information of the obstacle to be processed in the image coordinate system using the intrinsic parameters of the camera.

[0016] In an optional embodiment, when there are multiple pieces of second position information, determining a target image corresponding to the obstacle to be processed from the vehicle surrounding environment image based on the second position information includes:

[0017] Based on each piece of second position information, an obstacle image corresponding to the second position information is extracted from the vehicle surrounding environment image containing the second position information; an obstacle image with the largest size is determined from all obstacle images, and the obstacle image with the largest size is determined as the target image.

[0018] The obstacle attribute identification method provided in this embodiment determines the largest obstacle image among all obstacle images as the target image, so that when the pedestrian false alarm model is used to identify the target image later, more and finer image features can be obtained, thereby improving the accuracy of the identification result.

[0019] In an optional embodiment, after inputting the target image into a pre-built pedestrian false alarm model to obtain a discrimination result, the method further includes:

[0020] When the judgment result is that the obstacle to be processed is not a pedestrian, the obstacle to be processed is filtered.

[0021] The obstacle attribute determination method provided in this embodiment directly filters out obstacles to be processed whose obstacle attributes are invalid, thereby avoiding incorrect braking during driving and improving driving stability and safety.

[0022] In an optional embodiment, after inputting the target image into a pre-built pedestrian false alarm model to obtain a discrimination result, the method further includes:

[0023] When the result of the judgment is that the obstacle to be processed is a pedestrian, the obstacle to be processed is retained.

[0024] In a second aspect, the present invention provides a device for determining obstacle attributes, comprising:

[0025] An acquisition module is configured to obtain first position information of an obstacle to be processed, where the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, whose obstacle attribute is a pedestrian and whose attribute confidence level is lower than a preset threshold. A conversion module is configured to map the first position information to an image of the vehicle surrounding environment captured by a camera mounted on the vehicle body based on internal and external parameters of the camera, thereby obtaining second position information of the obstacle to be processed in the image of the vehicle surrounding environment. An extraction module is configured to determine a target image corresponding to the obstacle to be processed from the image of the vehicle surrounding environment based on the second position information. A discrimination module is configured to input the target image into a pre-built pedestrian false alarm model, which outputs a discrimination result. When the discrimination result corresponds to the obstacle attribute, the obstacle attribute is determined to be valid.

[0026] In a third aspect, the present invention provides a computer device, comprising:

[0027] The memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for determining the attributes of an obstacle according to any one of the embodiments of the first aspect by executing the computer instructions.

[0028] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to enable a computer to execute the method for determining the attributes of an obstacle of any embodiment of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0030] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0031] Figure 1 Schematic diagram of the flow of the method for determining obstacle attributes according to an embodiment of the present disclosure;

[0032] Figure 2 Schematic diagram of the flow of the method for determining obstacle attributes according to an embodiment of the present disclosure;

[0033] Figure 3 Schematic diagram of the flow of the method for determining obstacle attributes according to an embodiment of the present disclosure;

[0034] Figure 4 Schematic diagram of the structure of the obstacle attribute determination device according to an embodiment of the present disclosure;

[0035] Figure 5 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features therein can be combined with each other in the absence of conflict.

[0037] The following description sets forth many specific details to facilitate a full understanding of the present disclosure. However, the present disclosure may also be implemented in other ways than those described herein. It is apparent that the embodiments described in the specification are only some of the embodiments of the present disclosure, not all of them. All other embodiments derived by those skilled in the art based on the embodiments of the present disclosure without inventive effort are also within the scope of protection of the present disclosure.

[0038] During autonomous driving, obstacles (such as weeds or trees growing in roadside grass) may sometimes be mistakenly identified as pedestrians. In this case, the vehicle will trigger emergency braking, resulting in traffic accidents such as rear-end collisions, affecting driving safety. To avoid misidentifying obstacles as pedestrians, the current method mainly uses a preset confidence threshold to determine whether the obstacle attribute is a pedestrian. For example, if obstacle A is identified as a pedestrian and its confidence level is lower than the confidence threshold, the system determines that obstacle A is not a pedestrian and ignores obstacle A, controlling the vehicle to continue driving. However, the reason for the low confidence level of obstacle A may be that the characteristics of obstacle A were not fully recognized during the recognition process. In this case, if the confidence level is used alone as the basis for determining misidentification, obstacle A will be ignored, resulting in serious traffic accidents and casualties. To this end, the present invention provides a method, device, computer equipment and medium for determining the attributes of obstacles to achieve accurate determination of obstacle attributes.

[0039] According to an embodiment of the present invention, an embodiment of a method for determining obstacle attributes is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0040] In this embodiment, a method for determining the attributes of an obstacle is provided, which can be used in a vehicle control device. Figure 1 FIG. 1 is a flow chart of a method for determining obstacle attributes according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0041] Step S101: obtaining first position information of an obstacle to be processed, wherein the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, the obstacle attribute is a pedestrian, and the attribute confidence is lower than a preset threshold.

[0042] Specifically, whether an obstacle is within a preset range centered on the vehicle can be determined based on the obstacle's location information. The obstacle's location information, attributes, and attribute confidence can be obtained from the obstacle's description. This description is derived from the identification and analysis of the vehicle's surrounding point cloud and surrounding environment imagery, describing the obstacle's characteristics. It should be noted that the preset range is the minimum safe distance for emergency braking.

[0043] For example, the first position information of the obstacle to be processed can be the position information of the obstacle to be processed in the vehicle coordinate system, such as the coordinates corresponding to the obstacle center point; or the position information of the three-dimensional detection frame corresponding to the obstacle to be processed in the vehicle coordinate system, such as the coordinates of each vertex of the three-dimensional detection frame, or the coordinates of the center point of the three-dimensional detection frame and the length, width, and height information of the three-dimensional detection frame. The vehicle coordinate system is a three-dimensional coordinate system with the vehicle as the coordinate origin.

[0044] Step S102 : Based on the internal and external parameters of the camera mounted on the vehicle body, the first position information is mapped to the vehicle surrounding environment image captured by the camera to obtain the second position information of the obstacle to be processed in the vehicle surrounding environment image.

[0045] Specifically, the camera's internal and external parameters are used to map the first position of the obstacle to be processed in the 3D vehicle coordinate system to the image of the vehicle's surroundings, thereby obtaining the second position of the obstacle to be processed in the 2D image coordinate system. It should be noted that the first position information corresponds to the second position information. If the first position information is the position information of the obstacle to be processed relative to the 3D frame, the second position information is the position information of the obstacle to be processed relative to the 2D frame.

[0046] Specifically, the vehicle can have one or more cameras mounted on the vehicle. While the vehicle is in motion, each camera captures images of the vehicle's surroundings at the same frequency, and each camera has different internal and external parameters. Therefore, when multiple cameras are mounted on the vehicle, the first position information can be mapped to the images of the vehicle's surroundings captured by the corresponding camera using the corresponding internal and external parameters to obtain the corresponding second position information. However, due to the limited field of view of each camera, the number of second position information pieces does not match the number of cameras.

[0047] For example, in this embodiment, the vehicle body is equipped with four cameras, among which camera 1 is located in front of the vehicle, camera 2 is located in front of the left side of the vehicle, camera 3 is located in front of the right side of the vehicle, and camera 4 is located at the rear. The position of the obstacle M to be processed is within the field of view of camera 1 and camera 2. When the first position information of M is mapped using the internal and external parameters of camera 1, the second position information of M in the vehicle surrounding environment image captured by camera 1 can be obtained; since the obstacle M to be processed is also within the field of view of camera 2, the second position information of M in the vehicle surrounding environment image captured by camera 2 can also be obtained. However, since M is not within the field of view of camera 3 and camera 4, the second position information of M in the vehicle surrounding environment images captured by camera 3 and camera 4 respectively cannot be obtained. It can be seen that the number of second position information is not consistent with the number of cameras, but it can be determined that there is at least one second position information.

[0048] Step S103 : Based on the second position information, a target image corresponding to the obstacle to be processed is determined from the vehicle surrounding environment image.

[0049] Specifically, based on the second position information, the location of the obstacle to be processed can be determined from the vehicle surrounding environment image, and the obstacle to be processed at that location can be extracted from the vehicle surrounding environment image. The extracted obstacle to be processed is the target image. The extraction method includes but is not limited to cutout.

[0050] For example, if there is only one second position information obtained in step S102 , the obstacle to be processed corresponding to the second position information is determined from the vehicle surrounding environment image, and the obstacle image corresponding to the obstacle to be processed is determined as the target image.

[0051] For example, if there are multiple pieces of second information obtained through step S102, based on each piece of second position information, obstacle images corresponding to the second position information are extracted from the vehicle surrounding environment image containing the second position information; the largest obstacle image is determined from all obstacle images, and the largest obstacle image is determined as the target image.

[0052] For example, still taking the embodiment in step S102 as an example, after step S102, the second position information of the obstacle M to be processed in the vehicle surrounding environment image captured by the camera 1 is obtained, which is recorded as L 21 , and the second position information of the obstacle M to be processed in the vehicle surrounding environment image collected by camera 2 is also obtained, which is recorded as L 22 According to L 21 , the vehicle surrounding environment image captured by camera 1 (i.e., including the second position information L 21 The obstacle image M corresponding to the obstacle M to be processed is extracted from the vehicle surrounding environment image) 21 , according to L 22, the vehicle surrounding environment image captured by camera 2 (i.e., including the second position information L 22 The obstacle image M corresponding to the obstacle M to be processed is extracted from the vehicle surrounding environment image) 22 , from M 21 and M 22 Filter out the largest obstacle image M 22 , the largest obstacle image M 22 Determine the target image.

[0053] Step S104: input the target image into a pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs a discrimination result.

[0054] Specifically, the target image is input into the pedestrian false alarm model, which extracts features from the target image and analyzes and processes the extracted features to determine whether the target image has pedestrian characteristics. The model then outputs a judgment result, which can be either a pedestrian or a non-pedestrian. When the judgment result corresponds to the obstacle attributes, that is, the judgment result and the obstacle attributes are consistent, the obstacle attributes are determined to be valid. For example, if the judgment result is a pedestrian, the obstacle attributes are also pedestrian, which means that the judgment result corresponds to the obstacle attributes.

[0055] The obstacle attribute determination method provided in this embodiment, after obtaining first position information of an obstacle to be processed, first maps the first position information to an image of the vehicle surrounding environment captured by the camera using the internal and external parameters of the vehicle-mounted camera, thereby obtaining second position information of the obstacle to be processed in the image of the vehicle surrounding environment. Then, using the second position information, a target image corresponding to the obstacle is determined from the image of the vehicle surrounding environment. Finally, the target image is input into a pedestrian false alarm model to determine the validity of the obstacle attributes. This embodiment performs a secondary attribute determination on obstacles with pedestrian attributes within a preset range of the driving vehicle and with a confidence level below a preset threshold, namely, obstacles to be processed. The pedestrian false alarm model is used to perform more detailed reading and analysis of the features of the target image, thereby obtaining a more accurate determination result, thereby resolving the defect of misidentification and improving the stability and safety of the driving process.

[0056] In this embodiment, a method for determining the attributes of an obstacle is provided, which can be used in a vehicle control device. Figure 2 FIG. 1 is a flow chart of a method for determining obstacle attributes according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0057] Step S201 : obtaining first position information of an obstacle to be processed, wherein the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, the obstacle attribute is a pedestrian, and the attribute confidence is lower than a preset threshold.

[0058] Specifically, the above step S201 includes:

[0059] Step S2011: Obtain description information of the target obstacle, where the description information includes the location information, obstacle attributes, and attribute confidence of the obstacle.

[0060] In some optional implementations, step S2011 includes:

[0061] Step a1: Acquire an image of the vehicle's surrounding environment.

[0062] Specifically, the vehicle surrounding environment image is the image collected by each camera installed on the vehicle body at the current moment. The number of vehicle surrounding environment images is the same as the number of installed cameras, and due to the different camera installation positions, the camera's shooting field of view is different. Therefore, the vehicle surrounding environment images collected by each camera are also different.

[0063] In step a2, the vehicle surrounding environment image is input into a pre-built bird's-eye view prediction model, and the description information of the obstacle is output.

[0064] Specifically, an image of the vehicle's surroundings is fed into a bird's-eye view prediction model, which extracts and analyzes information from the input image to generate a 3D bird's-eye view in the vehicle's coordinate system. This 3D bird's-eye view includes each identified obstacle, its corresponding 3D detection frame, and its corresponding description. Because the vehicle's surroundings are captured while the vehicle is in motion, the speed at which it moves can easily lead to multiple obstacles being identified as a single obstacle. Therefore, the number of obstacles in the 3D bird's-eye view must be greater than or equal to the number of obstacles in the actual driving environment at the moment.

[0065] Step a3: Based on the description information of the obstacles, the obstacles are post-processed using a non-maximum suppression algorithm to determine the target obstacle from the obstacles and obtain the description information of the target obstacle.

[0066] Specifically, based on the obstacle description information, the non-maximum suppression (NMS) algorithm is used to post-process the pending obstacles to filter out the target obstacles. For example, if obstacle A in the actual environment is identified as three obstacles A1, A2, and A3 in the 3D bird's-eye view image, after post-processing using the NMS algorithm, A3 is selected as the obstacle corresponding to the optimal bounding box and identified as the target obstacle. It is important to emphasize that after post-processing, the number of target obstacles obtained is less than the number of obstacles in the original 3D bird's-eye view image, and greater than or equal to the number of obstacles in the actual driving environment.

[0067] The obstacle attribute determination method provided in this embodiment uses the NMS algorithm to post-process obstacles, remove some redundant obstacles, and screen out target obstacles with the best bounding box, thereby reducing the amount of obstacle data to a certain extent and improving work efficiency.

[0068] In step S2012, target obstacles within a preset range centered on the driving vehicle, whose obstacle attributes are pedestrians and whose attribute confidence is less than a preset threshold, are determined as obstacles to be processed.

[0069] Specifically, the system extracts location information, obstacle attributes, and attribute confidence from the target obstacle's description. Target obstacles within a preset range, centered on the driving vehicle, whose obstacle attributes are pedestrians and whose attribute confidence is less than a preset threshold, are identified as obstacles to be processed. Attribute confidence is the probability that an obstacle belongs to a certain attribute. A higher confidence level indicates a greater probability of the obstacle belonging to a certain attribute, while a lower confidence level indicates a lower probability of the obstacle belonging to a certain attribute.

[0070] Step S2013: extracting position information from the description information of the obstacle to be processed, and using the position information as first position information.

[0071] Step S202: Based on the internal and external parameters of the vehicle body mounted camera, the first position information is mapped to the vehicle surrounding environment image captured by the camera to obtain the second position information of the obstacle to be processed in the vehicle surrounding environment image. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0072] Step S203: Based on the second position information, determine the target image corresponding to the obstacle to be processed from the vehicle surrounding environment image. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0073] Step S204: Input the target image into the pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs the discrimination result. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0074] In this embodiment, a method for determining the attributes of an obstacle is provided, which can be used in a vehicle control device. Figure 3 FIG. 1 is a flow chart of a method for determining obstacle attributes according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0075] Step S301: Obtain the first position information of the obstacle to be processed. The obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, whose obstacle attribute is a pedestrian and whose attribute confidence is lower than a preset threshold. Figure 2 Step S201 of the illustrated embodiment will not be described in detail here.

[0076] Step S302 : Based on the internal and external parameters of the camera mounted on the vehicle body, the first position information is mapped to the vehicle surrounding environment image captured by the camera to obtain the second position information of the obstacle to be processed in the vehicle surrounding environment image.

[0077] Specifically, the above step S302 includes:

[0078] Step S3021: Using the extrinsic parameters of the camera, the first position information is converted into third position information of the obstacle to be processed in the camera coordinate system.

[0079] Specifically, extrinsic parameters describe the camera's position and orientation in the world coordinate system. These parameters typically include a rotation matrix and a translation vector. These parameters may vary depending on the camera's position or capture time. Using these parameters, we can convert the first position information in the vehicle coordinate system into the third position information of the obstacle being processed in the camera coordinate system. The camera coordinate system is a three-dimensional coordinate system with the camera as its origin.

[0080] Step S3022: Using the intrinsic parameters of the camera, the third position information is converted into second position information of the obstacle to be processed in the image coordinate system.

[0081] Specifically, intrinsic parameters describe the internal properties of a camera, including focal length, principal point (optical center) coordinates, and distortion coefficients. These parameters are fixed for a specific camera model. Using these parameters, the third position of the obstacle to be processed in the camera coordinate system is converted to its second position in the two-dimensional image coordinate system.

[0082] Step S303: Based on the second position information, determine the target image corresponding to the obstacle to be processed from the vehicle surrounding environment image. Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0083] Step S304: Input the target image into the pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs the discrimination result. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0084] Step S305: When the result of determination is that the obstacle to be processed is not a pedestrian, the obstacle to be processed is filtered.

[0085] Specifically, when the pedestrian false alarm model determines that the attribute of the obstacle to be processed is not a pedestrian, it means that the obstacle attribute of the obstacle to be processed is invalid, that is, the obstacle to be processed may actually be other obstacles such as weeds or trees, but it is mistakenly identified as a pedestrian during recognition. Therefore, the obstacle attribute recognition is invalid at that moment. At this time, the obstacle to be processed can be directly filtered.

[0086] The obstacle attribute determination method provided in this embodiment directly filters out obstacles to be processed whose obstacle attributes are invalid, thereby avoiding incorrect braking during driving and improving driving stability and safety.

[0087] Step S306: When the result of determination is that the obstacle to be processed is a pedestrian, the obstacle to be processed is retained.

[0088] Specifically, when the pedestrian false alarm model determines that the attribute of the obstacle to be processed is a pedestrian, it means that the obstacle attribute of the obstacle to be processed is valid, but the attribute confidence calculation is incorrect. At this time, the obstacle to be processed needs to be retained so that the vehicle can brake when it drives to the obstacle to be processed.

[0089] This embodiment also provides a device for determining obstacle attributes, which is used to implement the above-mentioned embodiments and preferred embodiments. Details already described will not be repeated here. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0090] This embodiment provides a device for determining the attributes of obstacles, such as Figure 4 Shown, including:

[0091] The acquisition module 401 is used to obtain first position information of an obstacle to be processed. The obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, whose obstacle attribute is a pedestrian and whose attribute confidence is lower than a preset threshold.

[0092] The conversion module 402 is configured to map the first position information to the vehicle surrounding environment image captured by the camera based on the internal and external parameters of the vehicle body mounted camera, and obtain the second position information of the obstacle to be processed in the vehicle surrounding environment image.

[0093] The extraction module 403 is configured to determine a target image corresponding to the obstacle to be processed from the vehicle surrounding environment image based on the second position information.

[0094] The discrimination module 404 is used to input the target image into the pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs a discrimination result. When the discrimination result corresponds to the obstacle attribute, it is determined that the obstacle attribute is valid.

[0095] In some optional implementations, the acquisition module 401 includes:

[0096] The acquisition submodule is used to obtain descriptive information of the target obstacle, which includes the target obstacle's location information, obstacle attributes, and attribute confidence. The target obstacle is obtained by post-processing the obstacle using a non-maximum suppression algorithm. The first determination submodule is used to determine as an obstacle to be processed any target obstacle within a preset range centered on the driving vehicle, whose obstacle attribute is a pedestrian and whose attribute confidence is less than a preset threshold. The extraction submodule is used to extract location information from the descriptive information of the obstacle to be processed and use the location information as the first location information.

[0097] In some optional implementations, the acquisition submodule includes:

[0098] The acquisition unit is used to acquire an image of the vehicle's surrounding environment; the output unit is used to input the image of the vehicle's surrounding environment into a pre-built bird's-eye view prediction model and output description information of the obstacles; and the post-processing unit is used to post-process the obstacles using a non-maximum suppression algorithm based on the description information of the obstacles to determine the target obstacle from the obstacles and obtain the description information of the target obstacle.

[0099] In some optional implementations, the conversion module 402 includes:

[0100] The first conversion submodule is used to convert the first position information into the third position information of the obstacle to be processed in the camera coordinate system by using the external parameters of the camera; the second conversion submodule is used to convert the third position information into the second position information of the obstacle to be processed in the image coordinate system by using the intrinsic parameters of the camera.

[0101] In some optional implementations, when there are multiple pieces of second location information, the extraction module 403 includes:

[0102] The extraction submodule is used to extract, based on each piece of second position information, the obstacle image corresponding to the second position information from the vehicle surrounding environment image containing the second position information; the second determination submodule is used to determine the largest obstacle image from all obstacle images and determine the largest obstacle image as the target image.

[0103] In some optional implementations, after the determination module 404, the apparatus further includes:

[0104] The filtering module is used to filter the obstacle to be processed when the judgment result shows that the obstacle to be processed is not a pedestrian.

[0105] In some optional implementations, after the determination module 404, the apparatus further includes:

[0106] The retaining module is used to retain the obstacle to be processed when the judgment result shows that the obstacle to be processed is a pedestrian.

[0107] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0108] The obstacle attribute determination device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0109] The embodiment of the present invention also provides a computer device having the above Figure 4 The obstacle attribute determination device shown.

[0110] See also Figure 5 , Figure 5 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0111] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0112] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0113] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0114] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0115] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0116] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0117] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device that includes the element.

[0118] The above are merely specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not limited to these embodiments, but is to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining obstacle attributes, characterized in that: include: Obtaining first position information of an obstacle to be processed, wherein the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, the obstacle attribute is a pedestrian, and the attribute confidence level is lower than a preset threshold; Based on internal and external parameters of a vehicle-mounted camera, the first position information is mapped to an image of the vehicle surrounding environment captured by the camera to obtain second position information of the obstacle to be processed in the image of the vehicle surrounding environment; Based on the second position information, determining a target image corresponding to the obstacle to be processed from the vehicle surrounding environment image; The target image is input into a pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs a discrimination result. When the discrimination result corresponds to the obstacle attribute, it is determined that the obstacle attribute is valid.

2. The method according to claim 1, characterized in that The obtaining of first position information of the obstacle to be processed includes: Obtaining description information of a target obstacle, the description information including location information of the target obstacle, attributes of the obstacle, and confidence of the attributes, wherein the target obstacle is obtained by post-processing the obstacle using a non-maximum suppression algorithm; Determining the target obstacle within a preset range centered on the driving vehicle, whose obstacle attribute is a pedestrian and whose attribute confidence is less than a preset threshold, as the obstacle to be processed; The position information is extracted from the description information of the obstacle to be processed, and the position information is used as the first position information.

3. The method according to claim 2, characterized in that The obtaining of description information of the target obstacle includes: Acquiring the vehicle surrounding environment image; Inputting the vehicle surrounding environment image into a pre-built bird's-eye view prediction model and outputting description information of obstacles; Based on the description information of the obstacles, the obstacles are post-processed using a non-maximum suppression algorithm to determine the target obstacle from the obstacles and obtain the description information of the target obstacle.

4. The method according to any one of claims 1 to 3, characterized in that The method of mapping the first position information to an image of the vehicle surrounding environment captured by the camera based on the internal and external parameters of the vehicle body mounted camera to obtain second position information of the obstacle to be processed in the image of the vehicle surrounding environment includes: Using the extrinsic parameters of the camera, the first position information is converted into third position information of the obstacle to be processed in the camera coordinate system; The third position information is converted into second position information of the obstacle to be processed in the image coordinate system by using the intrinsic parameters of the camera.

5. The method according to any one of claims 1 to 3, characterized in that When there are multiple pieces of second position information, determining a target image corresponding to the obstacle to be processed from the vehicle surrounding environment image based on the second position information includes: extracting, based on each piece of the second position information, an obstacle image corresponding to the second position information from the vehicle surrounding environment image containing the second position information; An obstacle image with the largest size is determined from all the obstacle images, and the obstacle image with the largest size is determined as the target image.

6. The method according to claim 1, characterized in that After inputting the target image into the pre-built pedestrian false alarm model to obtain a discrimination result, the method further includes: When the judgment result is that the obstacle to be processed is not a pedestrian, the obstacle to be processed is filtered.

7. The method according to claim 1, characterized in that After inputting the target image into the pre-built pedestrian false alarm model to obtain a discrimination result, the method further includes: When the determination result is that the obstacle to be processed is a pedestrian, the obstacle to be processed is retained.

8. A device for determining obstacle attributes, characterized in that: include: an acquisition module, configured to acquire first position information of an obstacle to be processed, wherein the obstacle to be processed is an obstacle within a preset range centered on the driving vehicle, the obstacle attribute is a pedestrian, and the attribute confidence level is lower than a preset threshold; a conversion module, configured to map the first position information to an image of the vehicle surrounding environment captured by the camera based on internal and external parameters of the camera mounted on the vehicle body, to obtain second position information of the obstacle to be processed in the image of the vehicle surrounding environment; an extraction module, configured to determine, from the vehicle surrounding environment image, a target image corresponding to the obstacle to be processed based on the second position information; A discrimination module is used to input the target image into a pre-built pedestrian false alarm model, and the pre-built pedestrian false alarm model outputs a discrimination result. When the discrimination result corresponds to the obstacle attribute, it is determined that the obstacle attribute is valid.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the obstacle attribute determination method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the obstacle attribute determination method according to any one of claims 1 to 7.

Images