Human body attribute correction method, device and equipment, medium and vehicle
By identifying the child safety seat in the vehicle cockpit and determining whether the target human body is sitting on the seat and correcting its attribute results, the problem of mis-detection of human attributes caused by child safety seats is solved, and the recognition accuracy is improved.
Patent Information
- Application Number
- CN202311482602.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-08
- Publication Date
- 2025-05-09
AI Technical Summary
When there is a child safety seat in the cabin of the vehicle, the child's body displacement causes the torso length information captured by the camera to be enlarged, causing false detection of human attribute recognition and low accuracy.
By obtaining the image to be detected in the car, image recognition is performed to determine whether there is a child safety seat, calculate the intersection ratio of the seat detection frame and the human body detection frame, determine whether the target human body is sitting on the child safety seat, and when determining that it is sitting on the seat, the attribute result is corrected as a child.
It effectively avoids the problem that the torso length information is amplified when children sit in a child safety seat, and improves the accuracy of human attribute recognition.
Smart Images
Figure CN119964195A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of image recognition technology, and in particular to a method, device, equipment, medium and vehicle for correcting human attributes. Background Art
[0002] Human attributes are the main basis for identifying the human body. When identifying human attributes, the presence of a child safety seat inside the cabin is not taken into account. When the human body is inside the cabin of the vehicle and there is a child safety seat inside the cabin of the vehicle, since the child safety seat has a considerable thickness, when the child sits in the child safety seat, the child's body will be displaced upward and forward. Since the spatial position and angle of the camera are fixed, the child's body will be magnified in the picture captured by the camera, which will cause the child's corresponding torso length information to be magnified, resulting in false detection of human attributes and low accuracy in human attribute recognition. Therefore, when there is a child safety seat inside the cabin, how to correct the human attributes becomes a technical problem that needs to be solved urgently. Summary of the invention
[0003] In order to solve the above technical problems, the present disclosure provides a human attribute correction method, device, equipment, medium and vehicle.
[0004] A first aspect of an embodiment of the present disclosure provides a method for correcting human attributes, the method comprising:
[0005] Acquire the image to be detected inside the vehicle;
[0006] Performing image recognition on the image to be detected to obtain a first detection result, and determining whether there is a child safety seat in the image to be detected according to the first detection result, wherein the first detection result includes a detection frame corresponding to a target object in the image to be detected and a category of the target object;
[0007] When it is determined that a child safety seat exists in the image to be detected, respectively calculating an intersection-over-union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result;
[0008] When the intersection-over-union ratio is greater than or equal to a preset intersection-over-union ratio threshold, a target human body detection frame corresponding to the intersection-over-union ratio is determined;
[0009] Get the crotch point corresponding to the target person in the target person detection frame;
[0010] Based on the crotch point and the seat detection frame, determine whether the target person is sitting in the child safety seat;
[0011] When it is determined that the target person is sitting in the child safety seat, the attribute result of the target person is corrected to a child.
[0012] A second aspect of an embodiment of the present disclosure provides a human attribute correction device, the device comprising:
[0013] An image acquisition module, used to acquire the image to be detected inside the vehicle;
[0014] An image recognition module is used to perform image recognition on the image to be detected to obtain a first detection result, and determine whether a child safety seat exists in the image to be detected according to the first detection result, wherein the first detection result includes a detection frame corresponding to a target object in the image to be detected and a category of the target object;
[0015] an intersection-and-union ratio calculation module, for respectively calculating the intersection-and-union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result when it is determined that a child safety seat exists in the image to be detected;
[0016] A first determination module is used to determine a target human body detection frame corresponding to the intersection-and-union ratio when the intersection-and-union ratio is greater than or equal to a preset intersection-and-union ratio threshold;
[0017] A crotch point acquisition module, used to acquire the crotch point corresponding to the target human body in the target human body detection frame;
[0018] A second determination module is used to determine whether the target person is sitting on the child safety seat based on the crotch point and the seat detection frame;
[0019] The attribute correction module is used to correct the attribute result of the target human body to a child when it is determined that the target human body is sitting in a child safety seat.
[0020] A third aspect of an embodiment of the present disclosure provides an electronic device, the device comprising:
[0021] Memory;
[0022] Processor; and
[0023] A computer program, wherein the computer program is stored in a memory and is configured to be executed by a processor to implement the human attribute correction method of the first aspect as described above.
[0024] A fourth aspect of the embodiments of the present disclosure provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the human body attribute correction method as described in the first aspect above is implemented.
[0025] A fifth aspect of an embodiment of the present disclosure provides a vehicle, comprising the electronic device of the third aspect described above.
[0026] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:
[0027] The human attribute correction method, device, equipment, medium and vehicle provided by the embodiments of the present disclosure can obtain an image to be detected in the vehicle, perform image recognition on the image to be detected, obtain a first detection result, determine whether a child safety seat exists in the image to be detected based on the first detection result, and when it is determined that a child safety seat exists in the image to be detected, calculate the intersection and union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result respectively, and when the intersection and union ratio is greater than or equal to a preset intersection and union ratio threshold, determine a target human body detection frame corresponding to the intersection and union ratio, and obtain the crotch portion corresponding to the target human body in the target human body detection frame. Point, based on the crotch point and the seat detection frame, determine whether the target person is sitting in the child safety seat. When it is determined that the target person is sitting in the child safety seat, the attribute result of the target person is corrected to a child. Therefore, when there is a child safety seat in the cabin, the child safety seat can be identified, and it can be judged whether the target person is in the child safety seat. When the target person is sitting in the child safety seat, the attributes of the target person are corrected, thereby avoiding the misdetection of human attributes caused by the amplification of the torso length information due to the target person sitting in the child safety seat, thereby improving the accuracy of human attribute recognition. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] 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.
[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0030] Figure 1 is a flow chart of a human attribute correction method provided by an embodiment of the present disclosure;
[0031] Figure 2 is a flow chart of a method for determining a reliable crotch point provided by an embodiment of the present disclosure;
[0032] Figure 3 is a flow chart of a method for determining whether a target person is sitting in a child safety seat provided by an embodiment of the present disclosure;
[0033] Figure 4 is a structural schematic diagram of a human attribute correction device provided by an embodiment of the present disclosure;
[0034] Figure 5 It is a structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION
[0035] 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 in the embodiments can be combined with each other without conflict.
[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.
[0037] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.
[0038] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0039] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".
[0040] Figure 1 It is a flowchart of a human attribute correction method provided by an embodiment of the present disclosure. The method can be executed by a human attribute correction device. The human attribute correction device can be implemented in software and / or hardware. The human attribute correction device can be configured in an electronic device, such as a server or a terminal or a server cluster, wherein the terminal can specifically include a mobile phone, a computer or a tablet computer, a vehicle-mounted terminal, or any device that can be used to process the human attribute correction method.
[0041] like Figure 1 As shown, the human body attribute correction method provided by the embodiment of the present disclosure includes the following steps.
[0042] S110: Acquire an image to be detected inside the vehicle.
[0043] In the embodiment of the present disclosure, the electronic device can obtain the image to be detected inside the vehicle based on the image acquisition device.
[0044] The image to be detected may be an image of the interior of a vehicle cabin, wherein the image to be detected may include at least one target human body, for example, the target human body may be a passenger in the vehicle cabin.
[0045] The image acquisition device may be at least one camera installed inside the vehicle cabin.
[0046] In some embodiments of the present disclosure, the electronic device can control an image acquisition device installed in the vehicle to acquire the image to be detected in real time, and upload the acquired image to be detected to the electronic device based on the image acquisition device, so that the electronic device can obtain the image to be detected.
[0047] In some other embodiments of the present disclosure, after receiving the human attribute correction instruction, the electronic device may obtain the image to be detected corresponding to the human attribute correction instruction by loading from a local memory based on the human attribute correction instruction.
[0048] S120, performing image recognition on the image to be detected to obtain a first detection result, and determining whether there is a child safety seat in the image to be detected based on the first detection result, wherein the first detection result includes a detection frame corresponding to the target object in the image to be detected and a category of the target object.
[0049] In an embodiment of the present disclosure, after acquiring the image to be detected, the electronic device detects, classifies and identifies the image to be detected based on a preset machine learning model, thereby obtaining a first detection result, and determines whether there is a child safety seat in the image to be detected based on the first detection result.
[0050] In the embodiment of the present disclosure, the first detection result may include a detection frame of a target object in the image to be detected and a category of the target object. When there are multiple target objects in the image to be detected, a detection frame of each target object and a category of the target object are obtained respectively. The target object may include a person in the vehicle and a child safety seat.
[0051] Among them, the preset machine learning model can be a model obtained by training with sample images containing target human bodies, and is used for detecting, classifying and identifying attributes of target human bodies, such as a human body detection and classification network.
[0052] S130: When it is determined that a child safety seat exists in the image to be detected, respectively calculate an intersection-over-union ratio between a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result.
[0053] In the embodiment of the present disclosure, the human body detection frame can be understood as a detection frame corresponding to the person in the car when the target object in the image to be detected is the person in the car.
[0054] The intersection-to-union ratio refers to the ratio of the intersection and union of the seat detection frame corresponding to the child safety seat and the human body detection frame corresponding to each target human body. The intersection refers to the area of the area containing the seat detection frame and the human body detection frame, and the union refers to the area of the seat detection frame and the human body detection frame combined.
[0055] Specifically, when the electronic device determines that there is a child safety seat in the image to be detected, it calculates the intersection and union ratio of the seat detection frame and each human detection frame based on the coordinate information of the human detection frame in the first detection result and the coordinate information of the seat detection frame corresponding to the child safety seat. The specific implementation method of calculating the intersection and union ratio of the seat detection frame and each human detection frame is similar to the existing implementation method of calculating the intersection and union ratio of the detection frames of two objects, which will not be repeated here.
[0056] S140: When the IoU ratio is greater than or equal to a preset IoU ratio threshold, determine a target human body detection frame corresponding to the IoU ratio.
[0057] In the embodiment of the present disclosure, after obtaining the intersection-and-union ratio of the seat detection frame and each human body detection frame, the electronic device compares the intersection-and-union ratio with a preset intersection-and-union ratio threshold. When the intersection-and-union ratio is greater than or equal to the preset intersection-and-union ratio threshold, it indicates that the target human body may be sitting in the child safety seat, and the target human body detection frame corresponding to the intersection-and-union ratio is further determined. When the intersection-and-union ratio is less than the preset intersection-and-union ratio threshold, it indicates that the target human body cannot be sitting in the child safety seat, and the target human body is discarded.
[0058] S150: Obtain the crotch point corresponding to the target human body in the target human body detection frame.
[0059] In the embodiment of the present disclosure, after the electronic device determines the target human body detection frame according to the intersection-union ratio, it inputs the area image of the target human body corresponding to the target human body detection frame into a preset human body key point detection network, and the preset human body key point detection network identifies the key points of the human body, and then obtains the crotch point corresponding to the target human body.
[0060] Optionally, the human key point detection network may be a pre-trained network model for identifying and predicting key points of a target human body in an image to be detected.
[0061] S160: Determine whether the target person is sitting on the child safety seat based on the crotch point and the seat detection frame.
[0062] In the disclosed embodiment, after obtaining the crotch point corresponding to the target human body, the electronic device determines whether the crotch point is located within the seat detection frame based on the position of the crotch point and the coordinate information of the seat detection frame, and further determines whether the target human body is sitting on the child safety seat.
[0063] S170: When it is determined that the target person is sitting in the child safety seat, the attribute result of the target person is corrected to a child.
[0064] In the disclosed embodiment, when the electronic device determines that the crotch point is within the seat detection frame, it determines that the target person is sitting in a child safety seat, thereby indicating that the target person is a child, and therefore corrects the attribute result of the target person in the image to be detected to a child.
[0065] In some embodiments of the present disclosure, when it is determined that the target person is not sitting in a child safety seat, the effect of the child safety seat on the target person's attributes is not considered, the body attribute correction process is terminated, and the attribute results of the target person are not corrected.
[0066] In the disclosed embodiment, it is possible to obtain an image to be detected in a vehicle, perform image recognition on the image to be detected, obtain a first detection result, determine whether there is a child safety seat in the image to be detected according to the first detection result, and when it is determined that there is a child safety seat in the image to be detected, respectively calculate the intersection-and-union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result, when the intersection-and-union ratio is greater than or equal to a preset intersection-and-union ratio threshold, determine a target human body detection frame corresponding to the intersection-and-union ratio, obtain a crotch point corresponding to the target human body in the target human body detection frame, determine whether the target human body is sitting on the child safety seat based on the crotch point and the seat detection frame, and when it is determined that the target human body is sitting on the child safety seat, correct the attribute result of the target human body to a child, thereby, when there is a child safety seat in the cabin, the child safety seat can be identified, and it can be judged whether the target human body is on the child safety seat, and when the target human body is sitting on the child safety seat, the attributes of the target human body are corrected, thereby avoiding the misdetection of human body attributes caused by the amplification of torso length information due to the target human body sitting on the child safety seat, thereby improving the accuracy of human body attribute recognition.
[0067] Furthermore, when it is determined that the target person is sitting on the child safety seat, before correcting the attribute result of the target person to a child, the human attribute correction method may further include: acquiring the attribute result of the target person.
[0068] In the embodiment of the present disclosure, the specific implementation method of obtaining the attribute results of the target human body is similar to the existing method of obtaining the attribute results of the human body, and will not be described in detail here.
[0069] Based on the above-mentioned embodiment of the present invention, in S120, image recognition is performed on the image to be detected to obtain a first detection result, and whether there is a child safety seat in the image to be detected is determined according to the first detection result. It can specifically include: inputting the image to be detected into a human body detection and classification network, and performing image recognition on the image to be detected by the human body detection and classification network to obtain a first detection result, the first detection result including a detection frame corresponding to the target object in the image to be detected and the category of the target object; determining whether there is a child safety seat according to the category of the target object in the first detection result.
[0070] In some embodiments of the present disclosure, a human body detection classification network may include a first backbone network, a first shoulder network, and a first head network. Based on the first backbone network, the first shoulder network, and the first head network, backbone features, shoulder features, and head features are extracted from the image to be detected, respectively, to obtain a feature map corresponding to the image to be detected. The feature map after feature extraction is traversed in the form of candidate frames to obtain a detection frame corresponding to each target human body in the image to be detected, and attribute recognition is performed on the detection frame corresponding to each target human body to obtain a detection frame corresponding to each target human body in the image to be detected and basic attributes of the human body. At the same time, if a child safety seat is present, the detection frame corresponding to the child safety seat is output, which is equivalent to outputting the detection frame corresponding to each target object in the image to be detected and the category of the target object, where the category of the target object includes people and child safety seats, thereby obtaining a first detection result.
[0071] Among them, when training the human body detection and classification network, the training data used are image data containing human bodies and child safety seats, and image data containing only human bodies, and each image data is labeled, wherein the annotation includes the detection box and category of the target object (human body, child safety seat), so that the trained human body detection and classification network has the ability to recognize the detection box and label the category of the human body and child safety seat in the image. Among them, the specific implementation method of using training data to train the human body detection and classification network is similar to the training method of the existing machine learning model, which will not be repeated here.
[0072] Specifically, after the electronic device acquires the image to be detected, it first preprocesses the image to be detected, wherein the preprocessing may include image cropping, image normalization processing, image filtering processing, etc., and inputs the preprocessed image to be detected into the human body detection and classification network, and the human body detection and classification network performs feature extraction on the preprocessed image to be detected to obtain a feature map corresponding to the image to be detected, and then the human body detection and classification network performs child safety seat recognition, human body recognition and human body attribute recognition on the feature map to obtain a first detection result corresponding to the target human body in the image to be detected, and the first detection result includes a detection frame corresponding to the target object in the image to be detected and the category of the target object. When the category of the target object is a child safety seat, it is determined that a child safety seat exists in the image to be detected. Thus, the image to be detected is detected, classified and recognized based on the human body detection and classification network, thereby improving the accuracy of determining whether there is a child safety seat in the image to be detected.
[0073] Furthermore, obtaining the crotch point corresponding to the target human body in the target human body detection frame in S150 may specifically include: inputting the area image corresponding to the target human body detection frame into the human body key point detection network, and performing key point recognition on the area image by the human body key point detection network to obtain a second detection result, which includes the crotch point corresponding to the target human body.
[0074] In some embodiments of the present disclosure, after obtaining a detection frame corresponding to a target human body, the detection frame corresponding to the target human body is input into a human key point detection network, wherein the human key point detection network may include a second backbone network, a second shoulder network, and a second head network, and backbone features, shoulder features, and head features are extracted from the target human body in the detection frame based on the second backbone network, the second shoulder network, and the second head network, respectively, to obtain a feature map corresponding to the target human body in the detection frame, and then the key points of the target human body are identified, and the identification results are matched with preset key points to obtain a second detection result, wherein the second detection result includes human key point information corresponding to the target human body.
[0075] Specifically, after obtaining the first detection result, the electronic device segments the area image corresponding to the target human body detection frame from the image to be detected to obtain the area image corresponding to the target human body, and inputs the area image of the target human body into the human body key point detection network. The human body key point detection network performs human body key point recognition on the area image to obtain the second detection result, wherein the second detection result includes key point information corresponding to the target human body, and searches and obtains the crotch point corresponding to the target human body and the position information of the crotch point from the key point information, thereby improving the accuracy of the crotch point corresponding to the target human body.
[0076] In the embodiment of the present disclosure, the crotch point includes a left crotch point, a right crotch point and a lower crotch point. After S150, the human attribute correction method may further include: determining at least one credible crotch point among the left crotch point, the right crotch point and the lower crotch point.
[0077] Further, Figure 2 is a flow chart of a method for determining a reliable crotch point provided by an embodiment of the present disclosure. Figure 2 As shown, the specific method for determining at least one credible crotch point among the left crotch point, the right crotch point and the lower crotch point includes steps S210-S250.
[0078] S210, obtaining a first confidence level corresponding to the left crotch point, a second confidence level corresponding to the right crotch point, and a third confidence level corresponding to the lower crotch point.
[0079] In the disclosed embodiment, after obtaining the crotch point corresponding to the target human body, the electronic device inputs the left crotch point position information, the right crotch point position information and the lower crotch point position information into a preset crotch point confidence mapping function, respectively, performs crotch point confidence calculations, and obtains a first confidence corresponding to the left crotch point, a second confidence corresponding to the right crotch point and a third confidence corresponding to the lower crotch point.
[0080] Among them, the preset crotch point confidence mapping function is a pre-set function for mapping the crotch point position information to the crotch point confidence, which can be understood as a function determined based on test data and personnel experience to characterize the one-to-one correspondence between the crotch point position information and the crotch point confidence.
[0081] The preset crotch point confidence mapping function is a mapping function for calculating the crotch point confidence by calculating and statistically generating the key point area position of the human body, the human body torso length, the human body torso length kernel density, etc. in a mathematical and statistical manner based on the test data. The key point area position of the human body can be the area position where the crotch point is located, or the area position where the key points adjacent to the crotch point are located.
[0082] S220: Compare the first confidence level, the second confidence level, and the third confidence level with preset crotch point confidence level thresholds respectively.
[0083] In the embodiment of the present disclosure, the preset crotch point confidence threshold is a pre-set confidence threshold for determining whether the crotch point is a credible crotch point.
[0084] The preset crotch point confidence threshold may be set based on personal experience, or may be a preset confidence threshold based on specific needs of the user.
[0085] S230: When the first confidence level is greater than or equal to the preset crotch point confidence level threshold, determine that the left crotch point is a credible crotch point.
[0086] S240: When the second confidence level is greater than or equal to a preset crotch point confidence level threshold, determine that the right crotch point is a credible crotch point.
[0087] S250: When the third confidence level is greater than or equal to a preset crotch point confidence level threshold, determine the lower crotch point as a credible crotch point.
[0088] In the disclosed embodiment, the confidence level corresponding to the crotch point can be obtained based on the acquired position information of the crotch point (including the left crotch point, the right crotch point and the lower crotch point), and then whether the crotch point is a credible crotch point can be determined based on the confidence level. When the crotch point is a credible crotch point, whether the target human body is sitting in the child safety seat is determined based on the position of the credible crotch point, thereby improving the accuracy of the result of determining whether the target human body is sitting in the child safety seat.
[0089] Figure 3 It is a flow chart of a method for determining whether a target person is sitting on a child safety seat provided in an embodiment of the present disclosure. In S160, based on the crotch point and the seat detection frame, it is determined whether the target person is sitting on the child safety seat, and steps S310-S340 can be specifically performed.
[0090] S310, performing weighted average calculation on pixel coordinates corresponding to at least one credible crotch point to obtain target pixel coordinates.
[0091] S320: Match the target pixel coordinates with at least one pixel coordinate in the seat detection frame.
[0092] S330: When the target pixel coordinates are consistent with any pixel coordinates in the seat detection frame, it is determined that the target person is sitting on the child safety seat.
[0093] S340: When the target pixel coordinates are inconsistent with all pixel coordinates in the seat detection frame, it is determined that the target person is not sitting in the child safety seat.
[0094] In the disclosed embodiment, the target pixel coordinates can be obtained based on the pixel coordinates corresponding to at least one trusted crotch point, and then the target pixel coordinates can be matched with any pixel coordinate in the seat detection frame to determine whether the target person is sitting in a child safety seat, thereby improving the accuracy of the result of determining whether the target person is sitting in a child safety seat, and thereby improving the accuracy of the human body attribute results.
[0095] Figure 4: is a schematic diagram of the structure of a human attribute correction device provided in an embodiment of the present disclosure. The human attribute correction device in the embodiment of the present disclosure can be set in an electronic device, and the electronic device can be a server or a terminal or a server cluster, wherein the terminal can specifically include a mobile phone, a computer or a tablet computer, a vehicle terminal, or any device that can be used to process the human attribute correction method, etc., which is not limited here.
[0096] like Figure 4 As shown, the human attribute correction device 400 may include an image acquisition module 410, an image recognition module 420, an intersection-and-union ratio calculation module 430, a first determination module 440, a crotch point acquisition module 450, a second determination module 460 and an attribute correction module 470.
[0097] The image acquisition module 410 can be used to acquire the image to be detected inside the vehicle.
[0098] The image recognition module 420 can be used to perform image recognition on the image to be detected, obtain a first detection result, and determine whether there is a child safety seat in the image to be detected based on the first detection result, wherein the first detection result includes a detection frame corresponding to the target object in the image to be detected and the category of the target object.
[0099] The IoU calculation module 430 may be used to calculate the IoU between the seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result when it is determined that a child safety seat exists in the image to be detected.
[0100] The first determination module 440 may be configured to determine a target human body detection frame corresponding to the IoU when the IoU is greater than or equal to a preset IoU threshold.
[0101] The crotch point acquisition module 450 may be used to acquire the crotch point corresponding to the target human body in the target human body detection frame.
[0102] The second determination module 460 may be used to determine whether the target person is sitting on the child safety seat based on the crotch point and the seat detection frame.
[0103] The attribute correction module 470 may be used to correct the attribute result of the target person to a child when it is determined that the target person is sitting in a child safety seat.
[0104] In the disclosed embodiment, it is possible to obtain an image to be detected in a vehicle, perform image recognition on the image to be detected, obtain a first detection result, determine whether there is a child safety seat in the image to be detected according to the first detection result, and when it is determined that there is a child safety seat in the image to be detected, respectively calculate the intersection-and-union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result, when the intersection-and-union ratio is greater than or equal to a preset intersection-and-union ratio threshold, determine a target human body detection frame corresponding to the intersection-and-union ratio, obtain a crotch point corresponding to the target human body in the target human body detection frame, determine whether the target human body is sitting on the child safety seat based on the crotch point and the seat detection frame, and when it is determined that the target human body is sitting on the child safety seat, correct the attribute result of the target human body to a child, thereby, when there is a child safety seat in the cabin, the child safety seat can be identified, and it can be judged whether the target human body is on the child safety seat, and when the target human body is sitting on the child safety seat, the attributes of the target human body are corrected, thereby avoiding the misdetection of human body attributes caused by the amplification of torso length information due to the target human body sitting on the child safety seat, thereby improving the accuracy of human body attribute recognition.
[0105] In some embodiments of the present disclosure, the image recognition module 420 can be specifically used to input the image to be detected into a human body detection and classification network, and the human body detection and classification network performs image recognition on the image to be detected to obtain a first detection result; and determine whether there is a child safety seat based on the category of the target object in the first detection result.
[0106] In some embodiments of the present disclosure, the crotch point acquisition module 450 can be specifically used to input the area image corresponding to the target human body detection frame into the human body key point detection network, and the human body key point detection network performs key point recognition on the area image to obtain a second detection result, which includes the crotch point corresponding to the target human body.
[0107] In some embodiments of the present disclosure, the crotch point includes a left crotch point, a right crotch point and a lower crotch point.
[0108] The human attribute correction device 400 may further include a trusted crotch point determination module.
[0109] The credible crotch point determination module can be used to determine at least one credible crotch point among the left crotch point, the right crotch point and the lower crotch point after obtaining the crotch point corresponding to the target human body in the target human body detection frame.
[0110] In some embodiments of the present disclosure, the trusted crotch point determination module can be specifically used to obtain a first confidence corresponding to the left crotch point, a second confidence corresponding to the right crotch point, and a third confidence corresponding to the lower crotch point; compare the first confidence, the second confidence, and the third confidence with a preset crotch point confidence threshold, respectively; when the first confidence is greater than or equal to the preset crotch point confidence threshold, determine that the left crotch point is a trusted crotch point; when the second confidence is greater than or equal to the preset crotch point confidence threshold, determine that the right crotch point is a trusted crotch point; when the third confidence is greater than or equal to the preset crotch point confidence threshold, determine that the lower crotch point is a trusted crotch point.
[0111] In some embodiments of the present disclosure, the second determination module 460 can be specifically used to perform weighted averaging calculation on the pixel coordinates corresponding to at least one trusted crotch point to obtain the target pixel coordinates; match the target pixel coordinates with at least one pixel coordinate in the seat detection frame; when the target pixel coordinates are consistent with any pixel coordinate in the seat detection frame, determine that the target person is sitting on the child safety seat; when the target pixel coordinates are inconsistent with all pixel coordinates in the seat detection frame, determine that the target person is not sitting on the child safety seat.
[0112] It should be noted that Figure 4 The human attribute correction device 400 shown can execute each step in the above method embodiment and realize each process and effect in the above method embodiment, which will not be described in detail here.
[0113] Figure 5 A schematic structural diagram of an electronic device provided by an embodiment of the present disclosure is shown.
[0114] In the disclosed embodiment, Figure 5 The electronic device shown may be a server or a terminal or a server cluster, wherein the terminal may specifically include a mobile phone, a computer or a tablet computer, a vehicle-mounted terminal, or any device that can be used for the human attribute correction method, etc., and is not limited here.
[0115] like Figure 5 As shown, the electronic device may include a processor 510 and a memory 520 storing computer program instructions.
[0116] Specifically, the processor 510 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0117] The memory 520 may include a large capacity memory for information or instructions. By way of example and not limitation, the memory 520 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 520 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 520 may be inside or outside the integrated gateway device. In a particular embodiment, the memory 520 is a non-volatile solid-state memory. In a particular embodiment, the memory 520 includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (Electrically Erasable Programmable ROM, EEPROM), an electrically rewritable ROM (EAROM), or a flash memory, or a combination of two or more of these.
[0118] The processor 510 reads and executes the computer program instructions stored in the memory 520 to perform the steps of the human attribute correction method provided in the embodiment of the present disclosure.
[0119] In one example, the electronic device may further include a transceiver 530 and a bus 540. Figure 5 As shown, the processor 510, the memory 520 and the transceiver 530 are connected via a bus 540 and communicate with each other.
[0120] The bus 540 includes hardware, software, or both. For example, but not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a Memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, the bus 540 may include one or more buses.
[0121] The embodiment of the present disclosure further provides a computer-readable storage medium, which may store a computer program. When the computer program is executed by a processor, the processor implements the human body attribute correction method provided by the embodiment of the present disclosure.
[0122] The above-mentioned storage medium may, for example, include a memory 520 of computer program instructions, and the above-mentioned instructions may be executed by the processor 510 of the electronic device to complete the human attribute correction method provided in the embodiment of the present disclosure. Optionally, the storage medium may be a non-temporary computer-readable storage medium, for example, the non-temporary computer-readable storage medium may be a ROM, a random access memory (Random Access Memory, RAM), a compact disc read-only memory (Compact Disc ROM, CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.
[0123] The embodiments of the present disclosure also provide a vehicle, which includes electronic equipment and can implement the various processes and effects in the above embodiments of the present disclosure, which will not be elaborated here.
[0124] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0125] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be 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 will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for correcting human attributes, characterized in that: The method comprises: Acquire the image to be detected inside the vehicle; Performing image recognition on the image to be detected to obtain a first detection result, and determining whether there is a child safety seat in the image to be detected according to the first detection result, wherein the first detection result includes a detection frame corresponding to a target object in the image to be detected and a category of the target object; When it is determined that a child safety seat exists in the image to be detected, respectively calculating an intersection-over-union ratio between a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result; When the IoU ratio is greater than or equal to a preset IoU ratio threshold, determining a target human body detection frame corresponding to the IoU ratio; Obtain the crotch point corresponding to the target human body in the target human body detection frame; Based on the crotch point and the seat detection frame, determining whether the target person is sitting on the child safety seat; When it is determined that the target person is sitting on the child safety seat, the attribute result of the target person is corrected to a child.
2. The method according to claim 1, characterized in that The performing image recognition on the image to be detected to obtain a first detection result, and determining whether a child safety seat exists in the image to be detected according to the first detection result, comprises: Inputting the image to be detected into a human body detection classification network, and performing image recognition on the image to be detected by the human body detection classification network to obtain a first detection result; Determine whether a child safety seat exists according to the category of the target object in the first detection result.
3. The method according to claim 1, characterized in that The step of obtaining the crotch point corresponding to the target human body in the target human body detection frame comprises: The area image corresponding to the target human body detection frame is input into the human body key point detection network, and the human body key point detection network performs key point recognition on the area image to obtain a second detection result, which includes the crotch point corresponding to the target human body.
4. The method according to claim 1, characterized in that The crotch points include a left crotch point, a right crotch point and a lower crotch point; After obtaining the crotch point corresponding to the target human body in the target human body detection frame, the method further includes: At least one credible crotch point among the left crotch point, the right crotch point and the lower crotch point is determined.
5. The method according to claim 4, characterized in that The determining of at least one credible crotch point among the left crotch point, the right crotch point and the lower crotch point comprises: Obtaining a first confidence level corresponding to the left crotch point, a second confidence level corresponding to the right crotch point, and a third confidence level corresponding to the lower crotch point; Comparing the first confidence level, the second confidence level and the third confidence level with preset crotch point confidence level thresholds respectively; When the first confidence level is greater than or equal to the preset crotch point confidence level threshold, determining the left crotch point as a credible crotch point; When the second confidence level is greater than or equal to the preset crotch point confidence level threshold, determining the right crotch point as a credible crotch point; When the third confidence level is greater than or equal to the preset crotch point confidence level threshold, the lower crotch point is determined to be a credible crotch point.
6. The method according to claim 4, characterized in that The determining whether the target person is sitting on the child safety seat based on the crotch point and the seat detection frame includes: Performing weighted average calculation on the pixel coordinates corresponding to the at least one credible crotch point to obtain the target pixel coordinates; Matching the target pixel coordinates with at least one pixel coordinate in the seat detection frame; When the target pixel coordinates are consistent with any pixel coordinates in the seat detection frame, it is determined that the target person is sitting on the child safety seat; When the target pixel coordinates are inconsistent with all pixel coordinates in the seat detection frame, it is determined that the target person is not sitting on the child safety seat.
7. A human body attribute correction device, characterized in that: include: An image acquisition module, used to acquire the image to be detected inside the vehicle; an image recognition module, configured to perform image recognition on the image to be detected to obtain a first detection result, and determine whether there is a child safety seat in the image to be detected according to the first detection result, wherein the first detection result includes a detection frame corresponding to a target object in the image to be detected and a category of the target object; an intersection-and-union ratio calculation module, configured to calculate the intersection-and-union ratio of a seat detection frame corresponding to the child safety seat and each human body detection frame in the first detection result respectively when it is determined that a child safety seat exists in the image to be detected; A first determination module is used to determine a target human body detection frame corresponding to the intersection-and-union ratio when the intersection-and-union ratio is greater than or equal to a preset intersection-and-union ratio threshold; A crotch point acquisition module, used to acquire the crotch point corresponding to the target human body in the target human body detection frame; A second determination module, configured to determine whether the target person is sitting on the child safety seat based on the crotch point and the seat detection frame; The attribute correction module is used to correct the attribute result of the target human body to a child when it is determined that the target human body is sitting on the child safety seat.
8. An electronic device, characterized in that: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.
10. A vehicle, characterized in that: Comprising the electronic device as claimed in claim 8.