Combined target detection method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of automatic driving, in particular to a combined target detection method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: respectively acquiring a first detection result of an independent target in a to-be-processed image; obtaining an association feature between any two independent targets, and determining a combination relationship between the independent targets according to the association feature; and generating a second detection result of the combined target based on the first detection result of the independent targets with the target combination relationship, wherein the combined target is obtained by combining the independent targets with the target combination relationship. By adopting the method, the efficiency of detecting the combined target and the accuracy of the detection result can be improved.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a combined target detection method, apparatus, computer equipment, storage medium and computer program product. Background Art

[0002] Autonomous driving refers to the synergy of technologies such as artificial intelligence, visual computing, radar, monitoring devices and global positioning systems, which allow computers to automatically and safely operate motor vehicles without any active human operation. In autonomous driving, the detection of targets in the environment is a very important technical module. This technical module is based on computer vision technology and is also the core basic direction of computer vision research. Therefore, it has received widespread attention. Unlike image recognition, target detection not only needs to identify the objects in the image and give the corresponding categories, but also needs to give the location of the object in the form of a minimum bounding box. Depending on the different output results required for target detection, RGB images are generally used for target detection. The method of outputting object categories and the minimum bounding box on the image is called 2D target detection, while the detection of RGB images, RGB-D depth images and laser point clouds to output object categories and information such as length, width, height, and rotation angle in three-dimensional space is called 3D target detection.

[0003] In related technologies, most of the detection of 2D targets adopts a target detection algorithm based on candidate boxes. The image features are first extracted through a convolutional network, and then a fully connected network is used to generate candidate boxes based on the feature map. The candidate boxes are classified and the borders are regressed through the network to obtain the classification and regression results of the targets in the image.

[0004] However, the current target detection methods have the following technical problems:

[0005] In the process of target recognition, single target detection is achieved by extracting image features. However, when applied to the detection of complex targets, the detection efficiency and accuracy of the results are low. Summary of the invention

[0006] Based on this, it is necessary to provide a combined target detection method, device, computer equipment, computer-readable storage medium and computer program product that can improve the efficiency of combined target detection and the accuracy of detection results in response to the above technical problems.

[0007] In a first aspect, the present application provides a combined target detection method. The method comprises:

[0008] Respectively obtaining first detection results of independent targets in the image to be processed;

[0009] Acquire correlation features between any two of the independent targets, and determine a combination relationship between the independent targets according to the correlation features;

[0010] A second detection result for the combined target is generated based on the first detection result of the independent targets having a target combination relationship, where the combined target is obtained by combining the independent targets having a target combination relationship.

[0011] In one embodiment, the first detection result includes a first detection frame, and the acquiring of association features between any two of the independent targets and determining the independent targets having a combination relationship according to the association features includes:

[0012] Randomly pairing the independent targets, and respectively calculating the intersection and union parameters between the first detection frames corresponding to the two paired independent targets;

[0013] The pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold are screened out.

[0014] In one embodiment, after screening out the pairs of independent targets whose intersection parameter is greater than or equal to a preset combination threshold, the method further includes:

[0015] Mapping the selected pairs of independent targets into the same two-dimensional coordinate system based on the first detection result;

[0016] Relative position parameters between the independent targets are obtained, and a combination relationship between the independent targets is determined according to the relative position parameters.

[0017] In one embodiment, generating a second detection result of the combined target based on the first detection result of the independent targets having a combined relationship comprises:

[0018] Mapping the independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result;

[0019] A second detection frame for locating the combined target is generated, wherein the second detection frame encloses a first detection frame corresponding to the independent target.

[0020] In one embodiment, generating a second detection frame for locating the combined target, wherein the second detection frame encloses a first detection frame corresponding to the independent target, comprises:

[0021] generating a circumscribed detection frame surrounding the first detection frame associated with the combined target;

[0022] The external detection frame is expanded based on a preset margin algorithm to obtain the second detection frame.

[0023] In one embodiment, generating a second detection result for the combined target based on the first detection result of the independent targets having a combined relationship further includes:

[0024] Obtaining a first classification result of the independent targets having a combination relationship;

[0025] A second classification result of the combined target is determined according to the first classification result.

[0026] In a second aspect, the present application also provides a combined target detection device. The device comprises:

[0027] An independent target module, used to respectively obtain first detection results of independent targets in the image to be processed;

[0028] A combination screening module, used to obtain the correlation features between any two of the independent targets, and determine the combination relationship between the independent targets according to the correlation features;

[0029] A combined target module is used to generate a second detection result for the combined target based on the first detection result of the independent targets having a target combination relationship, and the combined target is obtained by combining the independent targets having a target combination relationship.

[0030] In one embodiment, the first detection result includes a first detection frame, and the combination screening module includes:

[0031] A target pairing module, used for randomly pairing the independent targets and respectively calculating the intersection and union parameters between the first detection frames corresponding to two pairs of the independent targets;

[0032] The pairing screening module is used to screen out the pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold.

[0033] In one embodiment, after the pairing screening module, it also includes:

[0034] A coordinate mapping module, used for mapping the selected pairs of independent targets into the same two-dimensional coordinate system based on the first detection result;

[0035] The relative position screening module is used to obtain relative position parameters between the independent targets and determine the combination relationship between the independent targets according to the relative position parameters.

[0036] In one embodiment, the combined target module includes:

[0037] A coordinate mapping module, used for mapping the independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result;

[0038] The detection frame determination module is used to generate a second detection frame for locating the combined target, wherein the second detection frame encloses the first detection frame corresponding to the independent target.

[0039] In one embodiment, the detection frame determination module includes:

[0040] A detection frame generation module, used to generate an external detection frame surrounding the first detection frame associated with the combined target;

[0041] The detection frame adjustment process is used to expand the external detection frame based on a preset margin algorithm to obtain the second detection frame.

[0042] In one embodiment, the combined target module further includes:

[0043] A first classification result module, used to obtain a first classification result of the independent targets having a combination relationship;

[0044] The second classification result module is used to determine the second classification result of the combined target according to the first classification result.

[0045] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps in the combined target detection method as described in any one of the embodiments in the first aspect are implemented.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in a combined target detection method as described in any one of the embodiments in the first aspect are implemented.

[0047] In a fifth aspect, the present application further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps in the combined target detection method as described in any one of the embodiments in the first aspect are implemented.

[0048] The above-mentioned combined target detection method, apparatus, computer device, storage medium and computer program product can achieve the following beneficial effects corresponding to the technical problems in the background technology by deducing the technical features in the exclusive rights:

[0049] When performing target detection analysis on an image to be processed, first obtain the first detection result of the independent target in the image to be processed according to the existing recognition algorithm, then calculate the correlation feature between any two independent targets according to the first detection result, and determine the combination relationship between the independent targets by screening the quantified correlation features, and finally combine the independent targets with the combination relationship for calculation, and generate the second detection result of the combined target after merging according to the first detection result, so as to realize the detection of the combined target based on the algorithm for independent target detection. In implementation, there is no need to pre-construct the complex features of the combined target and train the corresponding recognition model. It is only necessary to realize the detection of multiple independent targets in the combined target separately, screen and combine the detection results, and calculate the detection results of the screened combined target. The first detection result of the independent target can be used for analysis in the calculation, which helps to improve the initial construction cost of the combined target detection on the one hand, and helps to improve the efficiency and accuracy of the combined target detection on the other hand. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] FIG1 is a schematic diagram of a first flow chart of a combined target detection method in one embodiment;

[0051] Figure 2 is a second flow chart of a combined target detection method in another embodiment;

[0052] Figure 3 is a third flow chart of a combined target detection method in another embodiment;

[0053] Figure 4 is a schematic diagram of a fourth flow chart of a combined target detection method in another embodiment;

[0054] Figure 5 is a schematic diagram of a second detection frame in a combined target detection method in an embodiment;

[0055] Figure 6 is a fifth flow chart of a combined target detection method in another embodiment;

[0056] Figure 7 is a sixth flow chart of a combined target detection method in another embodiment;

[0057] Figure 8 is a structural block diagram of a combined target detection device in one embodiment;

[0058] FIG. 9 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0060] In related technologies, most of the detection of 2D targets adopts a target detection algorithm based on candidate boxes. The image features are first extracted through a convolutional network, and then a fully connected network is used to generate candidate boxes based on the feature map. The candidate boxes are classified and the borders are regressed through the network to obtain the classification and regression results of the targets in the image.

[0061] However, the current target detection methods have the following technical problems:

[0062] In the process of target recognition, single target detection is achieved by extracting image features. However, when applied to the detection of complex targets, the detection efficiency and accuracy of the results are low.

[0063] Based on this, an embodiment of the present application provides a combined target detection method.

[0064] In one embodiment, as shown in FIG1 , a combined target detection method is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0065] Step 102: Obtain first detection results of independent targets in the image to be processed respectively.

[0066] The image to be processed may refer to a two-dimensional image of the space to be detected acquired by the terminal through a visual sensor, and the image to be processed may also be imported from the outside into the terminal for processing. An independent target may refer to an object that is identified and analyzed as a whole in a conventional target detection scenario, such as a pedestrian, a vehicle, a roadblock, a lane line, etc. The first detection result may refer to the detection result of an independent target of the terminal, which may include a detection frame of the independent target, basic information identified for the independent target, etc.

[0067] Exemplarily, the terminal can obtain the image to be processed, and obtain the first detection result of the independent target in the image to be processed through the target detection algorithm. At this time, the terminal can obtain information of several independent targets in the image to be processed, such as motor vehicles, pedestrians, non-motor vehicles, etc.

[0068] Step 104: Obtain association features between any two of the independent targets, and determine a combination relationship between the independent targets based on the association features.

[0069] The association feature may refer to a feature index used to describe the correlation between two independent targets. The combination relationship may refer to the association relationship between independent targets, for example, there is no combination relationship between motor vehicles and pedestrians, but there is a combination relationship between non-motor vehicles and cyclists.

[0070] Exemplarily, after obtaining the first detection result, the terminal can calculate the correlation features between any two independent targets in the image to be processed, and the detection results of the independent targets can be used as the quantitative basis of the correlation features in the calculation. In this way, the terminal can know whether there is a combination relationship between the independent targets through the calculated quantitative features. Generally, the stronger the correlation feature between the two independent targets, the greater the possibility of having a combination relationship.

[0071] Step 106: Generate a second detection result for the combined target based on the first detection result of the independent targets having a target combination relationship, where the combined target is obtained by combining the independent targets having a target combination relationship.

[0072] The combined target may refer to a whole obtained by combining independent targets with a combined relationship, for example, a non-motor vehicle riding target obtained by combining independent non-motor vehicle targets and pedestrian targets. The second detection result may refer to a detection result of the combined target in target detection, and may include a detection frame of the combined target, basic information of the combined target, etc.

[0073] Exemplarily, after screening out independent targets with a combination relationship, the terminal can combine the grouped independent targets to obtain basic information of the combined target, and generate a second detection result of the combined target through the first detection result. At this time, the second detection result is obtained based on the information of the first detection result of the independent target contained in the combined target, and in the process of generating the second detection result, necessary calculations, analysis, and information combination can be performed based on the first detection result.

[0074] In the above-mentioned combined target detection method, reasonable deduction is performed in combination with the technical features in the embodiment, which can achieve the following beneficial effects of solving the technical problems raised in the background technology:

[0075] When performing target detection analysis on an image to be processed, first obtain the first detection result of the independent target in the image to be processed according to the existing recognition algorithm, then calculate the correlation feature between any two independent targets according to the first detection result, and determine the combination relationship between the independent targets by screening the quantified correlation features, and finally combine the independent targets with the combination relationship for calculation, and generate the second detection result of the combined target after merging according to the first detection result, so as to realize the detection of the combined target based on the algorithm for independent target detection. In implementation, there is no need to pre-construct the complex features of the combined target and train the corresponding recognition model. It is only necessary to realize the detection of multiple independent targets in the combined target separately, screen and combine the detection results, and calculate the detection results of the screened combined target. The first detection result of the independent target can be used for analysis in the calculation, which helps to improve the initial construction cost of the combined target detection on the one hand, and helps to improve the efficiency and accuracy of the combined target detection on the other hand.

[0076] In one embodiment, the Figure 2 As shown, the first detection result may include a first detection frame, and step 104 includes:

[0077] Step 202: randomly pair the independent targets, and respectively calculate the intersection and union parameters between the first detection frames corresponding to the two paired independent targets.

[0078] The intersection parameter may refer to a parameter feature for describing the ratio of the intersection to the union of two bounding boxes.

[0079] Exemplarily, in order to determine which independent targets in the image to be processed can be combined for analysis, the terminal can randomly pair the independent targets in the image to be processed. Subsequently, the terminal can respectively calculate the intersection-and-union parameters of the first detection boxes corresponding to the two paired independent targets. Specifically, the intersection-and-union parameters can be calculated by calling the intersection-and-union ratio function (IOU).

[0080] Step 204: Filter out the pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold.

[0081] The combination threshold may refer to a screening threshold for screening the intersection parameter. The combination threshold may be preset by a technician and may be 40%, 50%, etc.

[0082] Exemplarily, the terminal may screen out independent targets whose intersection parameters are greater than or equal to the preset combination threshold through a preset combination threshold, thereby determining targets that can be subjected to combined analysis.

[0083] In this embodiment, screening independent targets by combining thresholds helps to determine targets that can be analyzed in combination, thereby improving analysis efficiency.

[0084] In one embodiment, the Figure 3 As shown, after step 204, the following may also be included:

[0085] Step 302: Mapping the selected pairs of independent targets into the same two-dimensional coordinate system based on the first detection result.

[0086] Exemplarily, the terminal may map the selected pairs of independent targets to the same two-dimensional coordinate system for analysis based on the first detection result, which helps to achieve consistency in the analysis of the first detection results of different independent targets.

[0087] Step 304: Obtain relative position parameters between the independent targets, and determine the combination relationship between the independent targets according to the relative position parameters.

[0088] The relative position parameter may refer to a parameter used to describe the relative position relationship between independent targets in the same coordinate system.

[0089] Exemplarily, the terminal may obtain relative position parameters between independent targets in the same coordinate system by calculation, and determine the combination relationship between the independent targets according to the relative position parameters.

[0090] The terminal can calculate the coordinate of the bottom edge of the pedestrian detection frame on the Y axis (image height) and the Y-axis coordinate of the bottom edge of the non-motor vehicle detection frame The difference , which can be expressed as follows:

[0091]

[0092] And through the difference The ratio of the height of the non-motor vehicle detection frame to the height of the non-motor vehicle detection frame can be calculated as follows:

[0093]

[0094] In this way, the terminal can determine the combination relationship between the two by the ratio of the difference between the detection frames of the independent targets in the Y-axis direction and the height of the detection frames.

[0095] Correspondingly, the terminal can also calculate the coordinate of the bottom edge of the pedestrian detection box on the X-axis (image width) and the coordinate of the bottom edge of the non-motor vehicle detection frame on the X axis The difference , which can be expressed as follows:

[0096]

[0097] And through the difference The ratio of the height of the non-motor vehicle detection frame to the height of the non-motor vehicle detection frame can be calculated as follows:

[0098]

[0099] In this way, the terminal can determine the combination relationship between the two by the ratio of the difference between the detection frames of the independent targets in the X-axis direction and the width of the detection frames.

[0100] As a supplement, the terminal can also calculate the width of the intersection and union ratio area of ​​the pedestrian detection frame and the non-motor vehicle detection frame The width of the non-motor vehicle detection frame The ratio can be expressed as follows:

[0101]

[0102] In this way, the terminal can determine the combination relationship between the two by the ratio of the width of the intersection area to the width of the independent target detection frame.

[0103] In this embodiment, the combination relationship between the two pairs of independent targets is determined by the relative position relationship between the two pairs, which helps to improve the accuracy of combination relationship detection.

[0104] In one embodiment, the Figure 4 As shown, step 106 includes:

[0105] Step 402: Mapping the independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result.

[0106] Exemplarily, the terminal may map independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result.

[0107] Step 404: Generate a second detection frame for locating the combined target, wherein the second detection frame encloses the first detection frame corresponding to the independent target.

[0108] For example, it can be Figure 5 As shown, the terminal can generate a second detection frame for locating the combined target based on the first detection frame of the independent target.

[0109] In this embodiment, using the detection frame as the basis for combined target detection helps to improve the detection efficiency of the combined target.

[0110] In one embodiment, the Figure 6 As shown, step 404 includes:

[0111] Step 602: Generate a circumscribed detection frame that surrounds the first detection frame associated with the combined target.

[0112] Exemplarily, the terminal may generate an external detection frame surrounded by a first detection frame associated with the combined target. The external detection frame may be an external rectangular frame or a polygonal detection frame external to the first detection frame.

[0113] Step 604: dilate the external detection frame based on a preset margin algorithm to obtain the second detection frame.

[0114] Exemplarily, the terminal may dilate the external detection frame based on a preset margin algorithm to obtain a second detection frame. The margin algorithm may refer to an algorithm used to control the margin between the second detection frame and the first detection frame. The dilation processing here may be positive dilation or negative dilation, i.e., shrinkage.

[0115] In this embodiment, the margin control algorithm helps to control the margin between the external detection frame and the first detection frame according to application requirements, thereby improving the generation accuracy and application flexibility of the second detection frame.

[0116] In one embodiment, the Figure 7 As shown, step 106 may include:

[0117] Step 702: Obtain a first classification result of the independent targets having a combination relationship.

[0118] The classification result may refer to the basic information about the target obtained in addition to the location of the target when detecting and identifying the target in the target detection application.

[0119] Exemplarily, the terminal may obtain a first classification result of independent targets having a combination relationship, such as pedestrians, non-motor vehicles, motor vehicles, and the like.

[0120] Step 704: Determine a second classification result of the combined target according to the first classification result.

[0121] Exemplarily, the terminal may determine the second classification result of the combined target based on the first classification result. For example, when the first classification results in the combined target are pedestrians and non-motor vehicles, the second classification result of the combined target may be a non-motor vehicle rider.

[0122] In this embodiment, by acquiring classification information of the combined target, the detection effect of the combined target can be improved.

[0123] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0124] Based on the same inventive concept, the embodiment of the present application also provides a combined target detection device for implementing the combined target detection method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the combined target detection device provided below can refer to the limitations of the combined target detection method above, and will not be repeated here.

[0125] In one embodiment, Figure 8 As shown, a combined target detection device is provided, comprising: an independent target module, a combined screening module and a combined target module, wherein:

[0126] An independent target module, used to respectively obtain first detection results of independent targets in the image to be processed;

[0127] A combination screening module, used to obtain the correlation features between any two of the independent targets, and determine the combination relationship between the independent targets according to the correlation features;

[0128] A combined target module is used to generate a second detection result for the combined target based on the first detection result of the independent targets having a target combination relationship, and the combined target is obtained by combining the independent targets having a target combination relationship.

[0129] In one embodiment, the first detection result includes a first detection frame, and the combination screening module includes:

[0130] A target pairing module, used for randomly pairing the independent targets and respectively calculating the intersection and union parameters between the first detection frames corresponding to two pairs of the independent targets;

[0131] The pairing screening module is used to screen out the pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold.

[0132] In one embodiment, after the pairing screening module, it also includes:

[0133] A coordinate mapping module, used for mapping the selected pairs of independent targets into the same two-dimensional coordinate system based on the first detection result;

[0134] The relative position screening module is used to obtain relative position parameters between the independent targets and determine the combination relationship between the independent targets according to the relative position parameters.

[0135] In one embodiment, the combined target module includes:

[0136] A coordinate mapping module, used for mapping the independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result;

[0137] The detection frame determination module is used to generate a second detection frame for locating the combined target, wherein the second detection frame encloses the first detection frame corresponding to the independent target.

[0138] In one embodiment, the detection frame determination module includes:

[0139] A detection frame generation module, used to generate an external detection frame surrounding the first detection frame associated with the combined target;

[0140] The detection frame adjustment process is used to expand the external detection frame based on a preset margin algorithm to obtain the second detection frame.

[0141] In one embodiment, the combined target module further includes:

[0142] A first classification result module, used to obtain a first classification result of the independent targets having a combination relationship;

[0143] The second classification result module is used to determine the second classification result of the combined target according to the first classification result.

[0144] Each module in the above-mentioned combined target detection device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module above.

[0145] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG9. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a combined target detection method is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0146] Those skilled in the art will understand that the structure shown in FIG. 9 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0147] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0148] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0149] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0150] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0151] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0152] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A combined target detection method, characterized in that: The method comprises: Respectively obtaining first detection results of independent targets in the image to be processed; Acquire correlation features between any two of the independent targets, and determine a combination relationship between the independent targets according to the correlation features; A second detection result for the combined target is generated based on the first detection result of the independent targets having a target combination relationship, where the combined target is obtained by combining the independent targets having a target combination relationship.

2. The method according to claim 1, characterized in that The first detection result includes a first detection frame, and the acquiring of association features between any two of the independent targets and determining the independent targets having a combination relationship according to the association features includes: Randomly pairing the independent targets, and respectively calculating the intersection and union parameters between the first detection frames corresponding to the two paired independent targets; The pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold are screened out.

3. The method according to claim 2, characterized in that After screening out the pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold, the method further includes: Mapping the selected pairs of independent targets into the same two-dimensional coordinate system based on the first detection result; Relative position parameters between the independent targets are obtained, and a combination relationship between the independent targets is determined according to the relative position parameters.

4. The method according to claim 1, characterized in that: The generating a second detection result of the combined target based on the first detection result of the independent targets having a target combination relationship comprises: Mapping the independent targets having a combination relationship into the same two-dimensional coordinate system according to the first detection result; A second detection frame for locating the combined target is generated, wherein the second detection frame encloses a first detection frame corresponding to the independent target.

5. The method according to claim 5, characterized in that: The generating of a second detection frame for locating the combined target, wherein the second detection frame encloses a first detection frame corresponding to the independent target, comprises: generating a circumscribed detection frame surrounding the first detection frame associated with the combined target; The external detection frame is expanded based on a preset margin algorithm to obtain the second detection frame.

6. The method according to claim 1, characterized in that The generating a second detection result of the combined target based on the first detection result of the independent targets having a combined relationship further includes: Obtaining a first classification result of the independent targets having a combination relationship; A second classification result of the combined target is determined according to the first classification result.

7. A combined target detection device, characterized in that: The device comprises: An independent target module, used to respectively obtain first detection results of independent targets in the image to be processed; A combination screening module, used to obtain the correlation features between any two of the independent targets, and determine the combination relationship between the independent targets according to the correlation features; A combined target module is used to generate a second detection result for the combined target based on the first detection result of the independent targets having a target combination relationship, and the combined target is obtained by combining the independent targets having a target combination relationship.

8. The device according to claim 7, characterized in that The combined screening module comprises: A target pairing module, used for randomly pairing the independent targets and respectively calculating the intersection and union parameters between the first detection frames corresponding to two pairs of the independent targets; The pairing screening module is used to screen out the pairs of independent targets whose intersection parameters are greater than or equal to a preset combination threshold.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images