Human body attribute correction method, device and equipment, medium and vehicle

Abstract

The invention relates to a human body attribute correction method and device, equipment, a medium and a vehicle, and the method comprises the steps: obtaining a to-be-corrected image sequence in the vehicle, and obtaining a human body attribute result corresponding to a target human body in a next frame of target image of the to-be-corrected image sequence, and the length of a human body trunk; when it is determined that the human body trunk length is the extreme trunk length, the number of continuous image frames corresponding to the extreme trunk length is determined; and when the number of the continuous image frames is greater than or equal to a preset extreme value continuous frame threshold value, giving a preset number of extreme value trunk length gradient correction frames to the to-be-corrected image sequence, and correcting the to-be-corrected image sequence based on the extreme value trunk length gradient correction frames and the human body attribute result. According to the method, the strong relevance between the trunk length information of the human body and the human body attributes can be considered, gradient correction is carried out on the human body attributes by applying the strong relevance between the trunk length information of the human body and the human body attributes, and the accuracy of human body attribute recognition is improved.

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 human bodies. The identification of human attributes is also crucial in real life. For example, when giving safety reminders to occupants based on their actions, it is necessary to judge the human attributes of the occupants, that is, whether they are adults or children, and then take corresponding safety reminders and other measures based on the judgment results and the actions of the occupants. In existing solutions, when identifying human attributes, image data containing the human body to be detected is usually input into a preset machine learning model, which extracts image features and identifies human key point information from the image data, and then judges the attributes of the human body to be detected based on the image features and human key point information to determine whether the human body to be detected is an adult or a child.

[0003] However, in the existing human attribute recognition scheme, although the key point information of the human body is taken into consideration, that is, the human body's torso length information is used as one of the bases for judging human attributes, the strong correlation between the human body's torso length information and human body attributes is not taken into account. Therefore, how to use the strong correlation between the human body's torso length information and human body attributes to correct human body attributes has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure provides a human attribute correction method, device, equipment, medium and vehicle.

[0005] A first aspect of an embodiment of the present disclosure provides a method for correcting human attributes, the method comprising:

[0006] Acquire a sequence of images to be corrected;

[0007] Obtaining the human attribute results corresponding to the target human body in the target image and the human body torso length corresponding to the target human body, wherein the target image is the next frame image of the image sequence to be corrected;

[0008] Comparing the human body torso length with a preset extreme torso length threshold, and determining whether the human body torso length is an extreme torso length based on the comparison result;

[0009] When determining that the human body trunk length is an extreme trunk length, determining the number of continuous image frames corresponding to the extreme trunk length;

[0010] When the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, a preset number of extreme trunk length gradient correction frames are assigned to the image sequence to be corrected so that the image sequence to be corrected enters a gradient correction state, and attribute correction is performed on the image sequence to be corrected based on the extreme trunk length gradient correction frames and the human body attribute results, and the final attribute results of the target human body in the image sequence to be corrected are determined.

[0011] A second aspect of an embodiment of the present disclosure provides a human attribute correction device, the device comprising:

[0012] An image acquisition module, used for acquiring a sequence of images to be corrected;

[0013] An attribute acquisition module is used to obtain a human attribute result corresponding to a target human body in a target image and a human torso length corresponding to the target human body in a target image, wherein the target image is a subsequent frame image of an image sequence to be corrected;

[0014] A first determination module is used to compare the human body trunk length with a preset extreme trunk length threshold, and determine whether the human body trunk length is an extreme trunk length based on the comparison result;

[0015] A second determination module is used to determine the number of continuous image frames corresponding to the extreme torso length when determining that the human torso length is the extreme torso length;

[0016] The attribute correction module is used to assign a preset number of extreme trunk length gradient correction frames to the image sequence to be corrected when the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, so that the image sequence to be corrected enters a gradient correction state, and performs attribute correction on the image sequence to be corrected based on the extreme trunk length gradient correction frames and human attribute results, and determines the final attribute results of the target human body in the image sequence to be corrected.

[0017] A third aspect of an embodiment of the present disclosure provides an electronic device, the device comprising:

[0018] Memory;

[0019] Processor; and

[0020] 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.

[0021] 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.

[0022] A fifth aspect of an embodiment of the present disclosure provides a vehicle, comprising the electronic device of the third aspect described above.

[0023] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0024] The human attribute correction method, device, equipment, medium and vehicle provided by the embodiments of the present disclosure can obtain a sequence of images to be corrected, obtain a human attribute result corresponding to a target human body in a target image and a human torso length corresponding to the target human body in the target image, wherein the target image is a subsequent frame of the sequence of images to be corrected, compare the human torso length with a preset extreme torso length threshold, determine whether the human torso length is an extreme torso length based on the comparison result, and when it is determined that the human torso length is an extreme torso length, determine the number of continuous image frames corresponding to the extreme torso length, and when the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, assign a preset number of extreme torso length gradual correction frames to the sequence of images to be corrected, so that the sequence of images to be corrected enters a gradual correction state, perform attribute correction on the sequence of images to be corrected based on the extreme torso length gradual correction frames and the human attribute result, and determine the final attribute result of the target human body in the sequence of images to be corrected. Thus, the strong correlation between the human torso length information and the human attributes can be taken into account, and the human attributes are gradually corrected by applying the strong correlation between the human torso length information and the human attributes, thereby improving the accuracy of human attribute recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] 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.

[0026] 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.

[0027] Figure 1 is a flow chart of a human attribute correction method provided by an embodiment of the present disclosure;

[0028] Figure 2 is a flow chart of a method for correcting an attribute of an image sequence to be corrected provided by an embodiment of the present disclosure;

[0029] Figure 3 is a structural schematic diagram of a human attribute correction device provided by an embodiment of the present disclosure;

[0030] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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".

[0036] 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.

[0037] like Figure 1 As shown, the human body attribute correction method provided by the embodiment of the present disclosure includes the following steps.

[0038] S110: Obtain a sequence of images to be corrected.

[0039] In the embodiment of the present disclosure, the electronic device may acquire a sequence of images to be corrected based on an image acquisition device.

[0040] The image sequence to be corrected may be an image sequence of the interior of a vehicle cabin, wherein each image to be corrected in the image sequence may include at least one target human body, for example, the target human body may be a passenger in the vehicle cabin, or may be an image sequence taken in any scene.

[0041] The image sequence to be corrected is a sequence of images that have been pre-collected and have undergone human attribute recognition and need to be corrected for human attribute. The image sequence to be corrected includes multiple images to be corrected and human attribute results corresponding to the target human body in each image to be corrected. The multiple images to be corrected in the image sequence to be corrected are sorted in time sequence.

[0042] The image sequence to be corrected may be an image sequence pre-stored in the electronic device, and the image sequence to be corrected may be acquired based on the identifier of the image sequence to be corrected.

[0043] The image acquisition device may be at least one camera installed inside the vehicle cabin.

[0044] Specifically, after receiving the human attribute correction instruction, the electronic device may obtain the to-be-corrected image sequence corresponding to the human attribute correction instruction by loading from the local memory based on the human attribute correction instruction.

[0045] S120, obtaining a human body attribute result corresponding to a target human body in a target image and a human body trunk length corresponding to the target human body, wherein the target image is a subsequent frame image of the image sequence to be corrected.

[0046] In an embodiment of the present disclosure, the electronic device acquires a target image based on an image acquisition device. After acquiring the target image, the electronic device performs detection classification and attribute recognition on the target image based on a preset machine learning model to obtain a detection frame of a target human body corresponding to the target image and a first detection result corresponding to the target human body, wherein the first detection result includes a basic adult confidence and a basic child confidence corresponding to the target human body and a detection frame corresponding to the target human body. The target human body is then identified at key points to obtain key point information of the target human body, and then a human trunk length corresponding to the target human body is determined based on the key point information of the human body. The first detection result and the human trunk length are then input into a preset human attribute mapping function for calculation to obtain a human attribute result corresponding to the target human body in the target image.

[0047] Optionally, the preset machine learning model can be a model obtained by training with sample images containing a target human body, and is used for detecting, classifying and identifying attributes of a target human body, such as a human body detection and classification network.

[0048] 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.

[0049] In the embodiment of the present disclosure, the human attribute results may include belonging to an adult and belonging to a child.

[0050] S130: Compare the human body trunk length with a preset extreme trunk length threshold, and determine whether the human body trunk length is an extreme trunk length based on the comparison result.

[0051] In the embodiment of the present disclosure, the preset extreme torso length threshold is a pre-set threshold for determining whether the torso length of a human body is an extreme torso length, wherein the preset extreme torso length threshold includes a preset maximum torso length threshold and a preset minimum torso length threshold.

[0052] Specifically, after obtaining the human body torso length, the electronic device compares the human body torso length with a preset extreme torso length threshold to determine whether the human body torso length meets the preset extreme torso length threshold, and then determines whether the human body torso length is the extreme torso length. When the preset extreme torso length threshold is met, the human body torso length is determined to be the extreme torso length.

[0053] S140. When it is determined that the human body trunk length is an extreme trunk length, determine the number of continuous image frames corresponding to the extreme trunk length.

[0054] In the embodiment of the present disclosure, the extreme torso length includes a maximum torso length and a minimum torso length.

[0055] The number of continuous image frames corresponding to the extreme trunk length can be understood as the number of image frames to be corrected that are located before the target image in the image sequence to be corrected, are continuously adjacent to the target image, and have the same extreme trunk length as the target image.

[0056] Exemplarily, taking the case where the extreme torso length is the maximum torso length as an example, the image sequence to be corrected contains a total of nine images, that is, the target image is the tenth image, and the torso length of the target person in the target image is the extreme torso length. At this time, determine whether the human torso length corresponding to the target person in the ninth image in the image sequence to be corrected that is located before the target image is the maximum torso length. When the human torso length corresponding to the target person in the ninth image is the maximum torso length, continue to determine the human torso length corresponding to the target person in the eighth image. When the human torso length corresponding to the target person in the eighth image is the maximum torso length, continue to determine the human torso length corresponding to the target person in the seventh image. If the human torso length corresponding to the target person in the seventh image is the minimum torso length or the non-extreme torso length, then determine that the number of continuous image frames corresponding to the extreme torso length is 2.

[0057] S150. When the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, a preset number of extreme trunk length gradient correction frames are assigned to the image sequence to be corrected so that the image sequence to be corrected enters a gradient correction state, and attribute correction is performed on the image sequence to be corrected based on the extreme trunk length gradient correction frames and the human body attribute results to determine the final attribute results of the target human body in the image sequence to be corrected.

[0058] In an embodiment of the present disclosure, after determining the number of continuous image frames corresponding to the extreme torso length, the electronic device compares the number of continuous image frames with a preset extreme continuous frame threshold. When the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, a preset number of extreme torso length gradient correction frames are assigned to the image sequence to be corrected so that the image sequence to be corrected enters a gradient correction state. Based on the extreme torso length gradient correction frames and the human body attribute results, attribute correction is performed on the image sequence to be corrected to determine the final attribute results of the target human body in the image sequence to be corrected.

[0059] Among them, performing attribute correction on the image sequence to be corrected based on the extreme torso length gradient correction frame and the human attribute result, and determining the final attribute result of the target human body in the image sequence to be corrected may include: when the human attribute result corresponding to the target human body in the target image is consistent with the attribute result corresponding to the extreme torso length, outputting the human attribute result of the target image, and voting for the target human body in the first image sequence in the image sequence to be corrected based on the human attribute result, to obtain the final attribute result of the target human body corresponding to the image sequence to be corrected; when the human attribute result corresponding to the target human body in the target image is inconsistent with the attribute result corresponding to the extreme torso length, reducing the number of extreme torso length gradient correction frames by 1, and after removing the human attribute result of the target image, voting for the target human body in the second image sequence in the image sequence to be corrected, to obtain the final attribute result corresponding to the target human body in the image sequence to be corrected.

[0060] In the embodiment of the present disclosure, the human attribute results may include belonging to an adult and belonging to a child.

[0061] In the disclosed embodiment, a sequence of images to be corrected can be obtained, a human body attribute result corresponding to a target human body in a target image and a human body trunk length corresponding to the target human body in the target image can be obtained, the target image is the next frame image of the sequence of images to be corrected, the human body trunk length is compared with a preset extreme trunk length threshold, and whether the human body trunk length is an extreme trunk length is determined based on the comparison result. When the human body trunk length is determined to be an extreme trunk length, the number of continuous image frames corresponding to the extreme trunk length is determined, and when the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, a preset number of extreme trunk length gradient correction frames are given to the sequence of images to be corrected, so that the sequence of images to be corrected enters a gradient correction state, attribute correction is performed on the sequence of images to be corrected based on the extreme trunk length gradient correction frames and the human body attribute result, and the final attribute result of the target human body in the sequence of images to be corrected is determined. Thus, the strong correlation between the human body trunk length information and the human body attributes can be taken into account, and the human body attributes are gradually corrected by applying the strong correlation between the human body trunk length information and the human body attributes, thereby improving the accuracy of human body attribute recognition.

[0062] On the basis of the above-mentioned embodiments of the present disclosure, obtaining the human attribute results corresponding to the target human body in the target image and the human torso length corresponding to the target human body in S120 may specifically include: performing image recognition on the target image based on a human detection classification network to obtain a first detection result, the first detection result including a detection frame corresponding to the target human body in the target image and a basic adult confidence and a basic child confidence corresponding to the target human body; inputting the detection frame area image corresponding to the target human body into a human key point detection network, performing key point detection on the target human body by the human key point detection network, and obtaining key point position information corresponding to the target human body; inputting the key point position information into a preset torso length mapping function to perform torso length calculation to obtain a torso length value corresponding to the target human body; inputting the basic adult confidence, the basic child confidence and the human torso length into a preset human attribute mapping function for calculation to obtain the human attribute results corresponding to the target human body in the target image.

[0063] In the embodiment of the present disclosure, the preset torso length mapping function is a pre-set function for mapping key point position information of a human body to torso length.

[0064] The preset human body attribute mapping function is a preset function used to map human body basic attribute information and human body trunk length information into human body attribute results.

[0065] The preset torso length mapping function and the preset human body attribute mapping function are respectively generated by calculating the human body torso length, the human body torso length kernel density, the human body basic confidence level, etc. based on the test data using mathematical statistics, and are used to calculate the human body torso length and human body attributes respectively.

[0066] 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 target image respectively to obtain a feature map corresponding to the target image. The feature map after feature extraction is traversed in a candidate frame manner to obtain a detection frame corresponding to each target human body in the target image, 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 target image and basic attributes of the human body (including basic adult confidence and basic child confidence). Then, the detection frame corresponding to the target human body is input into a human body key point detection network, wherein the human body key point detection network may include a second backbone network, a second shoulder network, and a second head network. Based on the second backbone network, the second shoulder network, and the second head network, backbone features, shoulder features, and head features are extracted from the target human body in the detection frame 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 the human body key point information corresponding to the target human body.

[0067] Specifically, after acquiring the target image, the electronic device first preprocesses the target image, wherein the preprocessing may include image cropping, image normalization processing, image filtering processing, etc., and inputs the preprocessed target image into the human body detection classification network, and the human body detection classification network performs feature extraction on the preprocessed target image to obtain a feature map corresponding to the target image, and then the human body detection classification network performs 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 target image, wherein the first detection result includes a detection frame of the target human body and basic human body attributes corresponding to the target human body (including Basic adult confidence and basic child confidence), and segment the regional image corresponding to the detection box corresponding to the target human body, that is, the regional image of the target human body, and input the regional image of the target human body into the human key point detection network. The human key point detection network performs human key point recognition on the regional image, and matches the recognition result with the preset key point to obtain a second detection result, wherein the second detection result includes the key point position information corresponding to the target human body, and further, the key point position information corresponding to the target human body is input into the preset torso length mapping function to calculate the torso length, so as to obtain the human torso length value corresponding to the target person.

[0068] In the embodiments of the present disclosure, basic human attributes (including basic adult confidence and basic child confidence) corresponding to the target human body can be obtained through a human body detection classification network, and key point information of the human body can be obtained through a human body key point detection network, and the key point information can be mapped to the human body trunk length based on a preset torso length mapping function, thereby improving the accuracy of the obtained basic adult confidence, basic child confidence and human body trunk length.

[0069] In the embodiment of the present disclosure, the preset extreme trunk length threshold includes a preset maximum trunk length threshold and a preset minimum trunk length threshold.

[0070] In S130, the human body torso length is compared with a preset extreme torso length threshold, and whether the human body torso length is an extreme torso length is determined based on the comparison result, which may specifically include: comparing the human body torso length with a preset maximum torso length threshold and a preset minimum torso length threshold, respectively; when the human body torso length is greater than the preset maximum torso length threshold, determining that the human body torso length is the maximum torso length; when the human body torso length is less than the preset minimum torso length threshold, determining that the human body torso length is the minimum torso length.

[0071] Further, when the human body trunk length is less than or equal to a preset maximum trunk length threshold and greater than or equal to a preset minimum trunk length threshold, the human body trunk length is determined to be a non-extreme trunk length.

[0072] Figure 2 is a flow chart of a method for correcting an attribute of an image sequence to be corrected provided by an embodiment of the present disclosure, such as Figure 2 As shown, the method for correcting the attributes of the image sequence to be corrected may specifically execute steps S210 - S290 .

[0073] S210: Determine whether the human body trunk length is an extreme trunk length.

[0074] In the disclosed embodiment, determining whether the human body trunk length is an extreme trunk length is similar to the specific implementation in the above embodiment, which will not be described in detail here.

[0075] In the disclosed embodiment, when the human body trunk length is an extreme trunk length, and the extreme trunk length is a maximum trunk length, steps S220-S250 are executed; when the extreme trunk length is a minimum trunk length, steps S260-S290 are executed.

[0076] S220. When the extreme trunk length is the maximum trunk length, determine the number of continuous image frames corresponding to the maximum trunk length.

[0077] In an embodiment of the present disclosure, determining the number of consecutive image frames corresponding to the extreme torso length includes: determining the first consecutive frame number in which the human torso length is the maximum torso length in the to-be-corrected images that are continuously adjacent to the target image in the to-be-corrected image sequence, and determining the first consecutive frame number as the number of consecutive image frames corresponding to the maximum torso length.

[0078] In some examples, continuous frames of maximum torso length are set to a, and continuous frames of minimum torso length are set to b. When there are five consecutive images to be corrected (the first to fifth images) in the image sequence to be corrected that are all of maximum torso length, a=5, b=0. When the sixth image is of minimum torso length, a is cleared to a=0, b=1, and so on.

[0079] Similarly, when the image sequence to be corrected contains ten images, starting from the tenth image, from the ninth image to the first image in chronological order, the number of continuous images that are continuously adjacent to the target image and in which the torso length of the target person in the target image is the maximum torso length, for example, if the torso length of the target person in the ninth image is the maximum torso length, then continue to determine that the torso length of the target person in the eighth, seventh, and sixth images are all the maximum torso length, and the torso length of the target person in the fifth image is the minimum torso length or a non-extreme torso length. At this time, the first continuous frame number of the maximum torso length corresponding to the torso length of the person in the target image is determined to be 4.

[0080] S230: When the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, output the human attribute result of the target image.

[0081] In the embodiment of the present disclosure, the preset extreme value continuous frame threshold is a preset continuous frame threshold used to determine whether to trigger a gradual correction state.

[0082] When the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, the maximum gradient correction state is entered. At this time, a preset number of maximum trunk length gradient correction frames is given to the image sequence to be corrected. For example, the maximum trunk length gradient correction frame number is set to m equal to t, where the preset number is t, and t is pre-set according to needs and can be modified or set arbitrarily. The preset number t can be understood as the preset continuous frame number of human attribute gradient correction.

[0083] S240: When the human body attribute result is an adult, voting is performed on the attributes of the target person in the first image sequence including the target image to determine a final attribute result of the target person.

[0084] In the embodiment of the present disclosure, when the human attribute result is an adult, it means that the human attribute result corresponding to the target human body in the target image is consistent with the human attribute result corresponding to the target human body in the image to be corrected in the continuous image frame number, then the attributes of the target person in the first image sequence containing the target image are directly voted to determine the final attribute result of the target person.

[0085] The first image sequence may be an image sequence determined according to a preset third preset number, wherein the first image sequence includes the target image.

[0086] For example, if the third preset number is 1000 frames, the first image sequence is an image sequence consisting of 1000 frames of images including the target image and closest to the target image. In some examples, the position of the target image in the image sequence is 1001 frames, and the first image sequence is from the 2nd frame to the 1001st frame; in other examples, the position of the target image in the image sequence is 500 frames, and the first image sequence is from the 1st frame to the 500th frame.

[0087] Furthermore, the specific implementation method of voting on the attributes of the target person in the first image sequence to determine the final attribute results of the target person is similar to the specific implementation method of voting on the attributes of the target person in the image sequence to be corrected based on the human attribute results corresponding to the target person in the target image and the first attribute results of each image to be corrected in the image sequence to be corrected when the human torso length is a non-extreme torso length, and a detailed description thereof will not be given here.

[0088] S250. When the human attribute result is a child, the number of maximum trunk length gradient correction frames is reduced by 1, and the attributes of the target person in the second image sequence except the target image are voted to determine the final attribute result of the target person.

[0089] In the disclosed embodiment, when the human attribute result is a child, it means that the human attribute result corresponding to the target human body in the target image is inconsistent with the human attribute result corresponding to the target human body in the image to be corrected in the corresponding continuous image frame number. Then, the number of maximum torso length gradient correction frames is directly reduced by 1, that is, m=t-1, and the attributes of the target person in the second image sequence except the target image are voted to determine the final attribute result of the target person.

[0090] The second image sequence may be an image sequence determined according to a preset third preset number, wherein the second image sequence does not include the target image.

[0091] For example, the third preset number is 1000 frames, and the position of the target image in the image sequence is frame 1001, then the second image sequence is frame 1 to frame 1000.

[0092] S260: When the extreme trunk length is the minimum trunk length, determine the number of continuous image frames corresponding to the minimum trunk length.

[0093] S270: When the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, output the human attribute result of the target image.

[0094] In the disclosed embodiment, when the extreme torso length is the minimum torso length, the number of continuous image frames corresponding to the minimum torso length is determined, which is similar to the above-mentioned determination of the number of continuous image frames corresponding to the maximum torso length when the extreme torso length is the maximum torso length, and will not be repeated here.

[0095] Among them, when the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, it enters the minimum gradient correction state. At this time, the image sequence to be corrected is given a preset number of minimum torso length gradient correction frames, for example, the number of minimum torso length gradient correction frames is set to n equal to t, where the preset number is t.

[0096] S280: When the human body attribute result is a child, voting is performed on the attributes of the target person in the first image sequence including the target image to determine a final attribute result of the target person.

[0097] In the embodiment of the present disclosure, when the human attribute result is a child, it means that the human attribute result corresponding to the target human body in the target image is consistent with the human attribute result corresponding to the target human body in the image to be corrected in the continuous image frame number. Then, the attributes of the target person in the first image sequence containing the target image are directly voted to determine the final attribute result of the target person.

[0098] S290. When the human attribute result is an adult, the number of minimum torso length gradient correction frames is reduced by 1, and the attributes of the target person in the second image sequence except the target image are voted to determine the final attribute result of the target person.

[0099] In the disclosed embodiment, when the human attribute result is an adult, it means that the human attribute result of the target image is inconsistent with the human attribute result corresponding to the target human body in the image to be corrected in the corresponding continuous image frame number, then the number of minimum torso length gradient correction frames is directly reduced by 1, that is, n=t-1, and the attributes of the target person in the second image sequence except the target image are voted to determine the final attribute result of the target person.

[0100] In the disclosed embodiments, when the image sequence to be corrected enters the gradual correction state, the attributes of the target human body in the image sequence to be corrected can be specifically gradual corrected according to the consistency between the human attribute results corresponding to the target human body in the target image and the first attribute results corresponding to the target person in the image to be corrected in the corresponding continuous image frames. That is, the strong correlation between the torso length information of the human body and the human attributes can be taken into consideration, and the strong correlation between the torso length information of the human body and the human attributes can be used to perform gradual correction on the human attributes, thereby improving the accuracy of human attribute recognition.

[0101] In some embodiments of the present disclosure, during the correction process of the image sequence to be corrected, when the human body trunk length is a non-extreme trunk length, the human body attribute correction method may also include: clearing the number of continuous image frames corresponding to the extreme trunk length determined according to the human body trunk length corresponding to the target human body in the image before the target image, and judging whether the image sequence to be corrected is in a gradual correction state; when it is determined that the image sequence to be corrected is not in a gradual correction state, outputting the human body attribute result corresponding to the target human body in the target image, and voting on the attributes of the target human body in the image sequence to be corrected based on the human body attribute result to determine the final attribute result of the target human body in the image sequence to be corrected; when it is determined that the image sequence to be corrected is in a gradual correction state, attribute correction is performed on the image sequence to be corrected based on the extreme trunk length gradual correction frame and the human body attribute result corresponding to the target human body in the target image to determine the final attribute result of the target human body in the image sequence to be corrected.

[0102] In the disclosed embodiment, the number of continuous image frames corresponding to the extreme torso length determined according to the human torso length corresponding to the target human body in the image before the target image is cleared, that is, the number of continuous image frames is initialized. For example, the continuous frames with the maximum torso length are set to a, and the continuous frames with the minimum torso length are set to b. At this time, a=3, b=0, then, a and b are directly initialized to 0, that is, a=b=0.

[0103] In some embodiments of the present disclosure, the electronic device determines the number of first images in which the target human body is an adult and the number of second images in which the target human body is a child in the target image and the image sequence to be corrected based on the human attribute results corresponding to the target human body in the target image and the attribute information of the target human body in the image sequence to be corrected; compares the number of first images with the number of second images, and when the number of first images is greater than or equal to the number of second images, determines that the human attribute results of the target human body are an adult, and when the number of second images is greater than the number of first images, determines that the human attribute results of the target human body are a child.

[0104] In some other embodiments of the present disclosure, the electronic device determines the number of first images in which the target human body is an adult and the number of second images in which the target human body is a child in the target image and the image sequence to be corrected based on the human attribute results corresponding to the target human body in the target image and the attribute information of the target human body in the image sequence to be corrected; compares the number of first images with the number of second images, and when the number of first images is greater than the number of second images, determines that the human attribute results of the target human body are an adult, and when the number of second images is greater than or equal to the number of first images, determines that the human attribute results of the target human body are a child.

[0105] In some other embodiments of the present disclosure, the electronic device determines the number of first images in which the target human body is an adult and the number of second images in which the target human body is a child in the target image and the image sequence to be corrected based on the human attribute results corresponding to the target human body in the target image and the attribute information of the target human body in the image sequence to be corrected; compares the number of first images with the number of second images, and when the number of first images is greater than the number of second images, determines that the human attribute results of the target human body are adult; when the number of second images is greater than the number of first images, determines that the human attribute results of the target human body are child; and when the number of first images is equal to the number of second images, does not output the human attribute results.

[0106] It should be noted that the specific implementation method of performing attribute correction on the image sequence to be corrected based on the extreme torso length gradient correction frame and the human attribute results corresponding to the target human body in the target image to determine the final attribute results of the target human body in the image sequence to be corrected is similar to the implementation method in the above-mentioned embodiment, that is, after the image sequence to be corrected enters the gradient correction state, the attribute correction is performed on the image sequence to be corrected based on the extreme torso length gradient correction frame and the human attribute results to determine the final attribute results of the target human body in the image sequence to be corrected, and will not be elaborated here.

[0107] In some embodiments of the present disclosure, when the number of continuous image frames is less than a preset extreme continuous frame threshold, the human attribute correction method may further include: determining whether the image sequence to be corrected is in a gradual correction state; when it is determined that the image sequence to be corrected is not in a gradual correction state, outputting the human attribute result corresponding to the target human body in the target image, and voting on the attributes of the target human body in the image sequence to be corrected based on the human attribute result, and determining the final attribute result of the target human body in the image sequence to be corrected; when it is determined that the image sequence to be corrected is in a gradual correction state, attribute correction is performed on the image sequence to be corrected based on the extreme torso length gradual correction frame and the human attribute result corresponding to the target human body in the target image, and determining the final attribute result of the target human body in the image sequence to be corrected. Among them, the specific implementation method is similar to the implementation method of the technical features corresponding to the above-mentioned human torso length being a non-extreme torso length, and will not be repeated here.

[0108] On the basis of the above-mentioned embodiments of the present disclosure, the gradient correction state includes a maximum gradient correction state and a minimum gradient correction state. The human body attribute correction method may further include: after the image sequence to be corrected enters the gradient correction state, based on the third human body attribute information of the image to be detected in the third image sequence located after the target image and the human body torso length of the target human body corresponding to the image to be detected in the third image sequence, refreshing the gradient correction state.

[0109] In the disclosed embodiment, the sequence to be detected may be a sequence of images acquired in real time, rather than a sequence of images of a fixed number. For example, the sequence of images to be corrected may be a sequence of images acquired in real time by a camera inside a vehicle. Therefore, as the number of images in the sequence of images to be corrected gradually increases, the torso length of the target human body in the obtained image to be detected will also change. After the sequence of images to be corrected enters the gradual correction state, the gradual correction state is refreshed according to the torso length of the target human body in the image to be detected after the target image, for example, from the maximum gradual correction state to the minimum gradual correction state, from the gradual correction state to the exit gradual correction state, etc.

[0110] The following is an example in which the human torso length of the target image is the maximum torso length.

[0111] In some embodiments of the present disclosure, when the image sequence to be corrected is currently in a maximum gradient correction state and the maximum trunk length gradient correction frame number is m=t (t is the preset continuous frame number of human attribute gradient correction), assuming t=100, then the minimum trunk length gradient correction frame number n=0, when the human attribute result corresponding to the target human body in multiple images to be detected in the subsequent third image sequence of the target image is a child, until the number m is reduced to 0, that is, m=0, then the gradient correction state is exited.

[0112] In other embodiments of the present disclosure, when the image sequence to be corrected is currently in a maximum gradient correction state and the maximum trunk length gradient correction frame number is m=50, and then the trunk lengths of the target human bodies in the subsequent multiple consecutive images to be detected are all minimum trunk lengths and meet the conditions for entering the minimum gradient correction state, then the gradient correction state of the image sequence to be corrected is refreshed from the maximum gradient correction state to the minimum gradient correction state, and the minimum trunk length gradient correction frame number is determined to be n=t.

[0113] Among them, the condition for entering the minimum gradient correction state is: the human body torso length corresponding to the target human body in the current image to be detected is the minimum torso length, and the number of continuous image frames corresponding to the minimum torso length is greater than or equal to the preset extreme continuous frame threshold.

[0114] In some other embodiments of the present disclosure, when the image sequence to be corrected is currently in a maximum gradual correction state and the maximum trunk length gradual correction frame number is m=50, and then the trunk lengths of the target human bodies in the subsequent multiple consecutive images to be detected are all maximum trunk lengths and meet the conditions for entering the maximum gradual correction state, then the maximum trunk length gradual correction frame number m in the image sequence to be corrected is refreshed to t, that is, m=t.

[0115] Among them, the condition for entering the maximum gradient correction state is: the human body torso length corresponding to the target human body in the current image to be detected is the maximum torso length, and the number of continuous image frames corresponding to the maximum torso length is greater than or equal to the preset extreme continuous frame threshold.

[0116] It should be noted that when the human torso length of the target image is the minimum torso length, the implementation method of refreshing the gradient correction state corresponding to the image sequence to be corrected is similar to the specific implementation method of refreshing the gradient correction state corresponding to the image sequence to be corrected when the human torso length is the minimum torso length, and will not be repeated here.

[0117] In the disclosed embodiment, after the image sequence to be corrected enters the gradient correction state, the gradient correction state can be refreshed in real time based on the third human attribute information of the image to be detected in the third image sequence located after the target image and the human torso length of the target human body corresponding to the image to be detected in the third image sequence, so as to ensure the accuracy of the human attribute results corresponding to the target human body in the image to be detected during the gradient correction process of the image sequence to be corrected.

[0118] Figure 3 : 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.

[0119] like Figure 3 As shown, the human attribute correction device 300 may include an image acquisition module 310 , an attribute acquisition module 320 , a first determination module 330 , a second determination module 340 and an attribute correction module 350 .

[0120] The image acquisition module 310 may be used to acquire a sequence of images to be corrected.

[0121] The attribute acquisition module 320 can be used to obtain the human attribute results corresponding to the target human body in the target image and the human body trunk length corresponding to the target human body, and the target image is the next frame image of the image sequence to be corrected.

[0122] The first determination module 330 may be configured to compare the human body trunk length with a preset extreme trunk length threshold, and determine whether the human body trunk length is an extreme trunk length based on the comparison result.

[0123] The second determination module 340 may be configured to determine the number of continuous image frames corresponding to the extreme torso length when determining that the human torso length is the extreme torso length.

[0124] The attribute correction module 350 can be used to assign a preset number of extreme torso length gradient correction frames to the image sequence to be corrected when the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, so that the image sequence to be corrected enters a gradient correction state, and perform attribute correction on the image sequence to be corrected based on the extreme torso length gradient correction frames and the human body attribute results, to determine the final attribute results of the target human body in the image sequence to be corrected.

[0125] In the disclosed embodiment, a sequence of images to be corrected can be obtained, and a human body attribute result corresponding to a target human body in a target image and a human body trunk length corresponding to the target human body can be obtained, the target image being the next frame image of the sequence of images to be corrected, the human body trunk length is compared with a preset extreme trunk length threshold, and whether the human body trunk length is an extreme trunk length is determined based on the comparison result. When it is determined that the human body trunk length is an extreme trunk length, the number of continuous image frames corresponding to the extreme trunk length is determined, and when the number of continuous image frames is greater than or equal to the preset extreme continuous frame threshold, a preset number of extreme trunk length gradient correction frames are given to the sequence of images to be corrected, so that the sequence of images to be corrected enters a gradient correction state, attribute correction is performed on the sequence of images to be corrected based on the extreme trunk length gradient correction frames and the human body attribute result, and the final attribute result of the target human body in the sequence of images to be corrected is determined. Thus, the strong correlation between the human body trunk length information and the human body attributes can be taken into account, and the human body attributes are gradiently corrected using the strong correlation between the human body trunk length information and the human body attributes, thereby improving the accuracy of human body attribute recognition.

[0126] In some embodiments of the present disclosure, the preset extreme trunk length thresholds include a preset maximum trunk length threshold and a preset minimum trunk length threshold.

[0127] The first determination module 330 can be specifically used to compare the human body trunk length with a preset maximum trunk length threshold and a preset minimum trunk length threshold respectively; when the human body trunk length is greater than the preset maximum trunk length threshold, the human body trunk length is determined to be the maximum trunk length; when the human body trunk length is less than the preset minimum trunk length threshold, the human body trunk length is determined to be the minimum trunk length.

[0128] In some embodiments of the present disclosure, the second determination module 340 can be specifically used to determine the first consecutive frame number in which the human body torso length is the maximum torso length in the to-be-corrected images continuously adjacent to the target image in the to-be-corrected image sequence when the extreme torso length is the maximum torso length, and determine the first consecutive frame number as the consecutive image frame number corresponding to the maximum torso length.

[0129] The attribute correction module 350 can be specifically used to, when the human body attribute result is an adult, vote on the attributes of the target human body in the first image sequence including the target image to determine the final attribute result of the target human body; when the human body attribute result is a child, reduce the number of maximum torso length gradient correction frames by 1, and vote on the attributes of the target human body in the second image sequence except the target image to determine the final attribute result of the target human body.

[0130] In some embodiments of the present disclosure, the second determination module 340 can be specifically used to determine, when the extreme torso length is the minimum torso length, the second consecutive frame number in which the human torso length is the minimum torso length in the to-be-corrected images continuously adjacent to the target image in the to-be-corrected image sequence, and determine the second consecutive frame number as the consecutive image frame number corresponding to the minimum torso length.

[0131] The attribute correction module 350 can be specifically used to, when the human body attribute result is a child, vote on the attributes of the target human body in the first image sequence including the target image to determine the final attribute result of the target human body; when the human body attribute result is an adult, reduce the number of minimum torso length gradient correction frames by 1, and vote on the attributes of the target human body in the second image sequence except the target image to determine the final attribute result of the target human body.

[0132] In some embodiments of the present disclosure, the human attribute correction device 300 may further include a third determination module.

[0133] The third determination module can be used to clear the number of continuous image frames corresponding to the extreme torso length determined according to the human torso length corresponding to the target human body in the image before the target image when the human torso length is a non-extreme torso length, and determine whether the image sequence to be corrected is in a gradual correction state; when it is determined that the image sequence to be corrected is not in a gradual correction state, output the human attribute result corresponding to the target human body in the target image, and vote on the attributes of the target human body in the image sequence to be corrected based on the human attribute result to determine the final attribute result of the target human body in the image sequence to be corrected; when it is determined that the image sequence to be corrected is in a gradual correction state, attribute correction is performed on the image sequence to be corrected based on the extreme torso length gradual correction frame and the human attribute result corresponding to the target human body in the target image, and the final attribute result of the target human body in the image sequence to be corrected is determined.

[0134] In some embodiments of the present disclosure, the gradient correction state includes a maximum gradient correction state and a minimum gradient correction state.

[0135] The human attribute correction device 300 may further include a gradual state refresh module.

[0136] The gradient state refresh module can be used to refresh the gradient correction state after the image sequence to be corrected enters the gradient correction state based on the third human body attribute information of the image to be detected in the third image sequence located after the target image and the human torso length of the target human body corresponding to the image to be detected in the third image sequence.

[0137] It should be noted that Figure 3 The human attribute correction device 300 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.

[0138] Figure 4 A schematic structural diagram of an electronic device provided by an embodiment of the present disclosure is shown.

[0139] In the disclosed embodiment, Figure 4 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.

[0140] like Figure 4 As shown, the electronic device may include a processor 410 and a memory 420 storing computer program instructions.

[0141] Specifically, the processor 410 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.

[0142] The memory 420 may include a large capacity memory for information or instructions. By way of example and not limitation, the memory 420 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 420 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 420 may be inside or outside the integrated gateway device. In a particular embodiment, the memory 420 is a non-volatile solid-state memory. In a particular embodiment, the memory 420 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.

[0143] The processor 410 reads and executes the computer program instructions stored in the memory 420 to perform the steps of the human attribute correction method provided in the embodiment of the present disclosure.

[0144] In one example, the electronic device may further include a transceiver 430 and a bus 440. Figure 4 As shown, the processor 410, the memory 420 and the transceiver 430 are connected via a bus 440 and communicate with each other.

[0145] The bus 440 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 440 may include one or more buses.

[0146] 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.

[0147] The above-mentioned storage medium may, for example, include a memory 420 of computer program instructions, and the above-mentioned instructions may be executed by the processor 410 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.

[0148] 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.

[0149] 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.

[0150] 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 a sequence of images to be corrected; Obtaining a human attribute result corresponding to a target human body in a target image and a human torso length corresponding to the target human body, wherein the target image is a subsequent frame image of the image sequence to be corrected; Comparing the human body trunk length with a preset extreme trunk length threshold, and determining whether the human body trunk length is an extreme trunk length based on the comparison result; When determining that the human body trunk length is an extreme trunk length, determining the number of continuous image frames corresponding to the extreme trunk length; When the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, a preset number of extreme trunk length gradient correction frames are assigned to the image sequence to be corrected so that the image sequence to be corrected enters a gradient correction state, and attribute correction is performed on the image sequence to be corrected based on the extreme trunk length gradient correction frames and the human body attribute results, and the final attribute results of the target human body in the image sequence to be corrected are determined.

2. The method according to claim 1, characterized in that The preset extreme trunk length thresholds include a preset maximum trunk length threshold and a preset minimum trunk length threshold; The step of comparing the human body trunk length with a preset extreme trunk length threshold, and determining whether the human body trunk length is an extreme trunk length based on the comparison result, comprises: Comparing the human body trunk length with the preset maximum trunk length threshold and the preset minimum trunk length threshold respectively; When the human body trunk length is greater than the preset maximum trunk length threshold, determining that the human body trunk length is the maximum trunk length; When the human body trunk length is less than the preset minimum trunk length threshold, it is determined that the human body trunk length is the minimum trunk length.

3. The method according to claim 1, characterized in that: When the extreme trunk length is the maximum trunk length, the extreme trunk length gradual correction frame is the maximum trunk length gradual correction frame; The step of determining the number of continuous image frames corresponding to the extreme trunk length comprises: Determine the first continuous frame number of the to-be-corrected images that are continuously adjacent to the target image in the to-be-corrected image sequence, in which the human torso length is the maximum torso length, and determine the first continuous frame number as the continuous image frame number corresponding to the maximum torso length; The correcting the image sequence to be corrected based on the extreme trunk length gradient correction frame and the human attribute result to determine the final attribute result of the target human body in the image sequence to be corrected includes: When the human body attribute result is an adult, voting on the attributes of the target human body in the first image sequence including the target image to determine a final attribute result of the target human body; When the human body attribute result is a child, the number of maximum trunk length gradient correction frames is reduced by 1, and the attributes of the target human body in the second image sequence except the target image are voted to determine the final attribute result of the target human body.

4. The method according to claim 1, characterized in that When the extreme trunk length is the minimum trunk length, the extreme trunk length gradual correction frame is the minimum trunk length gradual correction frame; The step of determining the number of continuous image frames corresponding to the extreme trunk length comprises: Determine a second continuous frame number in which a human body trunk length is a minimum trunk length in the image to be corrected that is continuously adjacent to the target image in the image sequence to be corrected, and determine the second continuous frame number as the continuous image frame number corresponding to the minimum trunk length; The correcting the image sequence to be corrected based on the extreme trunk length gradient correction frame and the human attribute result to determine the final attribute result of the target human body in the image sequence to be corrected includes: When the human body attribute result is a child, voting on the attributes of the target human body in the first image sequence including the target image to determine a final attribute result of the target human body; When the human attribute result is an adult, the number of the minimum trunk length gradual correction frames is reduced by 1, and the attributes of the target human body in the second image sequence except the target image are voted to determine the final attribute result of the target human body.

5. The method according to claim 1, characterized in that The method further comprises: When the human body torso length is a non-extreme torso length, the number of continuous image frames corresponding to the extreme torso length determined according to the human body torso length corresponding to the target human body in the image before the target image is cleared, and it is determined whether the image sequence to be corrected is in a gradual correction state; When it is determined that the image sequence to be corrected is not in a gradual correction state, outputting a human body attribute result corresponding to the target human body in the target image, and voting on the attributes of the target human body in the image sequence to be corrected based on the human body attribute result to determine a final attribute result of the target human body in the image sequence to be corrected; When it is determined that the image sequence to be corrected is in a gradual correction state, attribute correction is performed on the image sequence to be corrected based on the extreme torso length gradual correction frame and the human attribute result corresponding to the target human body in the target image to determine the final attribute result of the target human body in the image sequence to be corrected.

6. The method according to claim 1, characterized in that The gradual change correction state includes a maximum gradual change correction state and a minimum gradual change correction state, and the method further includes: After the image sequence to be corrected enters the gradual correction state, the gradual correction state is refreshed based on the third human attribute information of the image to be detected in the third image sequence located after the target image and the human torso length of the target human body corresponding to the image to be detected in the third image sequence.

7. A human body attribute correction device, characterized in that: include: An image acquisition module, used for acquiring a sequence of images to be corrected; An attribute acquisition module, used to acquire a human attribute result corresponding to a target human body in a target image and a human body trunk length corresponding to the target human body in the target image, wherein the target image is a subsequent frame image of the image sequence to be corrected; A first determination module is used to compare the human body trunk length with a preset extreme trunk length threshold, and determine whether the human body trunk length is an extreme trunk length based on the comparison result; A second determination module is used to determine the number of continuous image frames corresponding to the extreme torso length when determining that the human body torso length is an extreme torso length; The attribute correction module is used to assign a preset number of extreme trunk length gradient correction frames to the image sequence to be corrected when the number of continuous image frames is greater than or equal to a preset extreme continuous frame threshold, so that the image sequence to be corrected enters a gradient correction state, and perform attribute correction on the image sequence to be corrected based on the extreme trunk length gradient correction frames and the human attribute results, to determine the final attribute results of the target human body in the image sequence to be corrected.

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.

Images