A behavior recognition method, device, vehicle and storage medium

Abstract

The application provides a behavior recognition method and device, a vehicle and a storage medium. The method is applied to the field of vehicles and includes the following steps: acquiring a current frame image for behavior recognition; acquiring a first feature map set corresponding to a previous frame image of the current frame image; the first feature map set is a feature map set obtained by inputting the previous frame image into a convolution layer of a recognition network model for feature map fusion; inputting the current frame image and the first feature map set into the recognition network model for behavior recognition processing to obtain a behavior recognition result corresponding to the current frame image; if the number of continuous image frames subjected to the behavior recognition processing is greater than a preset frame number and the behavior recognition result is a preset behavior, the vehicle function corresponding to the preset behavior is started. The method can improve the convenience and accuracy of behavior recognition of the vehicle and the operability of starting the vehicle function by the user.

Description

Technical Field

[0001] This application relates to the field of vehicles, and more specifically, to a behavior recognition method, apparatus, vehicle, and storage medium in the field of vehicles. Background Technology

[0002] As vehicles become more and more common in daily life, the onboard functions that motor vehicles can perform are also becoming more and more numerous. In order to make it easier for users to activate various functions, the behavior of passengers in the vehicle can be identified, and the onboard functions can be activated based on the identification results.

[0003] In related technologies, when identifying passengers based on their behavior, the method used is to acquire data from each frame separately and then analyze the identification results of each frame in chronological order to obtain the passenger's behavior. However, this process is cumbersome and timing errors can lead to insufficient accuracy and efficiency. Summary of the Invention

[0004] This application provides a behavior recognition method, device, vehicle, and storage medium. The method can recognize feature map sets through a recognition network model, which avoids recognition errors caused by timing. It also reduces the complexity of data processing and improves data processing efficiency through feature map sets, thereby improving the convenience and accuracy of vehicle behavior recognition and the ease of operation for users to start vehicle functions.

[0005] Firstly, a behavior recognition method is provided, the method comprising:

[0006] Obtain the current frame image for behavior recognition, and obtain the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion.

[0007] The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image;

[0008] If the number of consecutive image frames for the behavior recognition process is greater than a preset number of frames, and the behavior recognition result is a preset behavior, then the vehicle is controlled to activate the vehicle function corresponding to the preset behavior.

[0009] The above technical solution enables the acquisition of a feature map set by progressively identifying multiple frames of images. A recognition network model is then used to perform behavior recognition on the current frame image and the corresponding feature map set, resulting in a behavior recognition result. The recognition network model then identifies the feature map set, thus avoiding recognition errors caused by timing issues. Furthermore, the feature map set reduces the complexity of data processing and improves data processing efficiency, thereby enhancing the convenience and accuracy of vehicle behavior recognition and the ease of operation for users to start vehicle functions.

[0010] In conjunction with the first aspect, in some possible implementations, the step of inputting the current frame image and the first feature map set into a recognition network model for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image includes:

[0011] The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the target behavior score of the preset behavior corresponding to the current frame image;

[0012] If the target behavior score is greater than a preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0013] Combining the first aspect and the above implementation methods, in some possible implementation methods, if the target behavior score is less than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be other behaviors, and the other behaviors are behaviors other than the preset behaviors.

[0014] In combination with the first aspect and the above implementation methods, in some possible implementation methods, after obtaining the current frame image for behavior recognition, the method further includes:

[0015] Based on the detection network model, human body parts are detected in the current frame image to determine the region image in the current frame image that contains the target body parts;

[0016] The image of the region is input into the recognition network model for part pose recognition processing to obtain the part pose recognition result;

[0017] If the position recognition result indicates that the target position is in a preset position, then the first memory value corresponding to the current frame image is set to the initial value;

[0018] If the position recognition result indicates that the target position is in another position, then the second memory value corresponding to the previous frame image is obtained, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image.

[0019] If the target behavior score is greater than a preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior, including:

[0020] If the first memory value is greater than the preset memory value, and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0021] The above technical solution can acquire and compare memory values ​​to ensure that the pose of the part corresponding to the current frame image is the preset pose, and the behavior recognition result is determined as the preset behavior based on the number of consecutive image frames, the first memory value and the target behavior score, thereby improving the accuracy of vehicle behavior recognition.

[0022] In combination with the first aspect and the above implementation methods, in some possible implementation methods, after the step of inputting the current frame image and the first feature map set into the recognition network model for behavior recognition processing, the method further includes:

[0023] If the number of consecutive image frames for the behavior recognition process is less than a preset number of frames, the behavior recognition result is discarded, and the process proceeds to the step of obtaining the current frame image for behavior recognition.

[0024] In combination with the first aspect and the above implementation methods, in some possible implementation methods, obtaining the first feature map set corresponding to the previous frame image of the current frame image includes:

[0025] If the current frame image is the first frame image, then an initial feature map set is obtained, and the feature value of each initial feature map in the initial feature map set is a preset feature value.

[0026] The above technical solution enables behavior recognition processing of the first frame image, thereby completing the behavior recognition process.

[0027] In combination with the first aspect and the above implementation methods, in some possible implementation methods, before acquiring the current frame image for behavior recognition, the method further includes:

[0028] Obtain the current scene state of the vehicle and match the current scene state with the parking scene state;

[0029] If the current scene state matches the parking scene state, then control the vehicle's camera to capture the current frame image.

[0030] The above technical solutions can prevent accidental activation of vehicle functions under certain conditions, thus avoiding driving accidents and improving vehicle safety and reliability.

[0031] Secondly, a behavior recognition device is provided, the device comprising:

[0032] The data acquisition unit is used to acquire the current frame image for behavior recognition and to acquire the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion.

[0033] The behavior recognition unit inputs the current frame image and the first feature map set into the recognition network model for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image;

[0034] The function activation unit is used to control the vehicle to activate the vehicle function corresponding to the preset behavior if the number of consecutive image frames for the behavior recognition processing is greater than a preset number of frames and the behavior recognition result is a preset behavior.

[0035] Thirdly, a vehicle is provided, including a memory for storing executable program code;

[0036] A processor is configured to call and run the executable program code from the memory, causing the vehicle to perform the method of the first aspect or any possible implementation thereof.

[0037] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0038] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof. Attached Figure Description

[0039] Figure 1 This is a system architecture diagram of a behavior recognition method provided in an embodiment of this application;

[0040] Figure 2 This is a flowchart illustrating a behavior recognition method provided in an embodiment of this application;

[0041] Figure 3 This is a flowchart illustrating a behavior recognition method provided in an embodiment of this application;

[0042] Figure 4 This is an example schematic diagram of feature map fusion provided in an embodiment of this application;

[0043] Figure 5This is an example schematic diagram of acquiring a region image provided in an embodiment of this application;

[0044] Figure 6 This is a schematic diagram of the structure of a behavior recognition device provided in an embodiment of this application;

[0045] Figure 7 This is a schematic diagram of the structure of a behavior recognition device provided in an embodiment of this application;

[0046] Figure 8 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application;

[0047] Figure 9 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application. Detailed Implementation

[0048] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0049] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0050] Figure 1 This is a system architecture diagram of a behavior recognition method provided in an embodiment of this application. For example... Figure 1 As shown, the behavior recognition method provided in this application embodiment can be applied to vehicles to realize the process of speed adjustment and gear shifting of vehicles. The system architecture provided in this specification embodiment mainly includes a vehicle 10, a vehicle-mounted system 20, and a camera 30. The vehicle 10 can be a motor vehicle, such as a fuel-powered vehicle, a hybrid vehicle, or an electric vehicle; the vehicle-mounted system 20 can be a computing terminal for data processing and system management in the vehicle, such as the vehicle's in-vehicle system; the vehicle-mounted system 20 includes a behavior recognition model 21 and vehicle functional components 22. The recognition network model 21 can be a network model with image recognition capabilities; the vehicle functional components 22 can be components in the vehicle used to perform vehicle functions, such as an audio controller, a lighting controller, etc.; the camera 30 can be a camera in the vehicle with image acquisition capabilities.

[0051] Specifically, the camera 30 in vehicle 10 captures images and transmits them to the vehicle infotainment system 20. The vehicle infotainment system 20 then performs behavior recognition on the images. In related technologies, the vehicle infotainment system 20 performs behavior recognition by sequentially recognizing multiple captured images according to their temporal order, and determines whether to control vehicle 10 to activate vehicle functions based on the recognition results. However, during the recognition process, timing errors can occur when recognizing images sequentially, and the data processing is cumbersome due to the simultaneous input of multiple images into the model, resulting in insufficient accuracy and efficiency in behavior recognition.

[0052] In this embodiment, the current frame image is captured by the camera 30 and transmitted to the vehicle system 20. The vehicle system 20 obtains the first feature map set corresponding to the previous frame image of the current frame image, and inputs the current frame image and the first feature map set into the recognition network model 21 for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image. If the number of consecutive image frames for behavior recognition processing is greater than the preset number of frames, and the behavior recognition result is the preset behavior, then the vehicle function component 22 in the vehicle system is transmitted with an instruction to control the various components in the vehicle 10 to start the vehicle function corresponding to the preset behavior. This improves the convenience and accuracy of vehicle behavior recognition, as well as the ease of operation for users to start vehicle functions.

[0053] based on Figure 1 The system architecture shown below will be combined with Figures 2-5 The behavior recognition method provided in the embodiments of this application will be described in detail.

[0054] Please see Figure 2 The diagram below illustrates a flowchart of a behavior recognition method provided in this application. Figure 2 As shown, the method in this application embodiment may include the following steps S101-S103.

[0055] S101, Obtain the current frame image for behavior recognition, and obtain the first feature map set corresponding to the previous frame image of the current frame image;

[0056] In one embodiment, when the vehicle meets the parking scenario state, in order to monitor in real time whether the driver has made a preset behavior that meets the vehicle function, it is necessary to acquire the current frame image for driver behavior recognition in real time, and acquire the first feature map set corresponding to the previous frame image of the current frame image.

[0057] The parking scenario can be a state where the vehicle is started and all internal components are in parking mode.

[0058] Vehicle functions can be functions pre-bound to preset behaviors, specifically controlling the operation modes of interior lights, audio systems, or sunroofs. Preset behaviors can be pre-programmed physical actions.

[0059] The current frame image can be the image captured by the vehicle's camera at the current moment.

[0060] The previous frame image can be an image that is adjacent to the current frame image in the image acquisition time sequence.

[0061] The first feature map set can be the set of feature maps obtained after feature map fusion in the convolutional layer of the recognition network model, which is input from the previous frame image. It can be understood that since each frame image includes the features of the previous frame image, the first feature map set can be considered to include the features of the previous frame image as well as the features corresponding to each image before the previous frame image.

[0062] A recognition network model can be a network model used to recognize images and determine whether a preset behavior exists in the image.

[0063] S102, input the current frame image and the first feature map set into the recognition network model for behavior recognition processing, so as to obtain the behavior recognition result corresponding to the current frame image;

[0064] In one embodiment, after obtaining the current frame image and the first feature map set, the current frame image and the first feature map set are output into the recognition network model. The recognition network model is used to perform behavior recognition processing based on the current frame image and the first feature map set to obtain the behavior recognition result corresponding to the current frame image.

[0065] Specifically, the method to obtain the behavior recognition result corresponding to the current frame image can be to use a recognition network model to perform behavior recognition processing on the current frame image and the first feature map set to obtain the target behavior score of the preset behavior corresponding to the current frame image. If the target behavior score is greater than the preset score threshold, the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0066] The target behavior score is the score obtained by the recognition network model for behavior recognition processing. It is used to indicate the probability that a preset behavior exists in the current frame image and the images before the current frame image. The higher the score value, the greater the probability that a preset behavior exists in the current frame image and the images before the current frame image.

[0067] The preset score threshold can be used to determine the behavior recognition result corresponding to the current frame image. The target behavior score is compared with the preset score threshold to determine the behavior recognition result corresponding to the current frame image.

[0068] The behavior recognition result can be determined by the target behavior score output by the recognition network model to determine the recognition result corresponding to the current frame image, which can include preset behaviors and other behaviors.

[0069] S103, if the number of consecutive image frames for behavior recognition processing is greater than the preset number of frames, and the behavior recognition result is the preset behavior, then control the vehicle to start the vehicle function corresponding to the preset behavior.

[0070] In one embodiment, after obtaining the recognition result of the recognition network model, the number of consecutive image frames that the recognition network model has performed behavior recognition processing is obtained, and the number of consecutive image frames is compared with a preset number of frames. If the number of consecutive image frames is greater than the preset number of frames and the behavior recognition result is a preset behavior, the vehicle is controlled to start the vehicle function corresponding to the preset behavior.

[0071] The number of consecutive image frames can be the number of frames that have undergone behavior recognition processing since the vehicle meets the parking scenario conditions.

[0072] The preset number of frames can be a pre-set number of frames that can meet the feature accumulation requirements of the initial stage of behavior recognition processing.

[0073] Specifically, if the number of consecutive image frames corresponding to the current frame image is greater than the preset number of frames, and the behavior recognition result corresponding to the current frame image is the preset behavior, then it can be determined that the driver has performed the preset behavior in front of the vehicle's camera, which meets the conditions for starting the vehicle's functions, and the vehicle is controlled to start the vehicle function corresponding to the preset behavior.

[0074] In this embodiment, a feature map set is obtained by acquiring multiple frames of images and performing step-by-step recognition. A recognition network model is then used to perform behavior recognition on the current frame image and the feature map set corresponding to the current frame image to obtain the behavior recognition result. The recognition network model recognizes the feature map set, which avoids recognition errors caused by time sequence and reduces the complexity of data processing by using the feature map set, thereby improving the data processing efficiency. This improves the convenience and accuracy of vehicle behavior recognition and the ease of operation for users to start vehicle functions.

[0075] The following will combine Figure 3 The present application provides a detailed description of a behavior recognition method through specific examples.

[0076] S201, Obtain the current scene state of the vehicle and match the current scene state with the parking scene state;

[0077] In one embodiment, the vehicle's infotainment system monitors various components of the vehicle in real time, obtains the current scene status of the vehicle based on the monitoring, and matches the current scene status with the parking scene status.

[0078] The various components of the vehicle can indicate the vehicle's gear position, starting status, and handbrake status, etc.

[0079] The current scene state can be the current state of the vehicle, such as the driving scene state or the parking scene state.

[0080] The parking scenario can be defined as a state where the vehicle is started and all internal components are in parking mode. Specifically, this could mean the vehicle is running, in park, and the handbrake is engaged. Understandably, if the vehicle uses a foot brake, it can be determined whether the foot brake is depressed.

[0081] S202, If the current scene state matches the parking scene state, control the vehicle's camera to capture the current frame image;

[0082] In one embodiment, if the current scene state matches the parking scene state, then the current scene state of the vehicle is determined to be the parking scene state, and the vehicle's camera is controlled to capture the current frame image.

[0083] The camera capturing the current frame image can be a camera used in the driver monitoring system to collect driver-related information. Because the driver monitoring system only monitors the driver, it avoids mis-identification based on the behavior of other passengers in the vehicle, which could lead to accidental vehicle activation.

[0084] Understandably, if the vehicle needs to activate a function in another scenario state, and / or needs to acquire the behavior of passengers other than the driver, then the current scenario state will be matched with the corresponding scenario state, and the current frame image will be acquired using either the driver monitoring system or the passenger monitoring system based on the matching result. Specific settings can be configured according to the actual situation.

[0085] S203, obtain the current frame image for behavior recognition, and obtain the first feature map set corresponding to the previous frame image of the current frame image;

[0086] In one embodiment, when the vehicle meets the parking scenario state, in order to monitor in real time whether the driver has made a preset behavior that meets the vehicle function, it is necessary to acquire the current frame image for driver behavior recognition in real time, and acquire the first feature map set corresponding to the previous frame image of the current frame image.

[0087] Vehicle functions can be pre-defined functions that are bound to preset behaviors. Specifically, these could be controlling the operation of interior lights, audio systems, or sunroofs to corresponding modes. Preset behaviors can be pre-programmed physical actions. For example, vehicle functions could control interior lights to a preset brightness, control the audio system to output music at a preset volume, or open the sunroof or its shutter.

[0088] The previous frame can be an image that is adjacent to the current frame in the image acquisition time sequence. For example, if the vehicle starts acquiring the current frame image at 12:00:00 and acquires 12 frames per second, and the current frame image is the 36th frame image, then the previous frame image is the 35th frame image.

[0089] The first feature map set can be the set of feature maps obtained after feature map fusion in the convolutional layer of the recognition network model, which is input from the previous frame image. It can be understood that since each frame image includes the features of the previous frame image, the first feature map set can be considered to include the features of the previous frame image as well as the features corresponding to each image before the previous frame image.

[0090] It should be noted that too many frames can lead to an overloaded feature map set, impacting data processing efficiency and accuracy. Therefore, a feasible approach is to use a duration-based sliding window with a length of 24 frames. In this case, the first feature map set includes features from the current frame and the 23 frames preceding it. For example, if the current frame is frame 36, the first feature map set would consist of the feature maps from frame 12 to frame 35 (a total of 23 frames). The specific duration of the sliding window can be adjusted based on actual conditions.

[0091] A recognition network model can be a network model used to recognize images and determine whether a preset behavior exists in the image.

[0092] Furthermore, after obtaining the feature maps corresponding to each target layer of the current frame image, each feature map is set as the first feature map set of the next frame of the current frame image and stored in the vehicle's in-vehicle system cache.

[0093] Furthermore, if the current frame image is the first frame image, then an initial feature map set is obtained and used as the first feature map set.

[0094] In this initial feature map set, each initial feature map has a preset feature value, for example, the preset feature value of each feature map is 0. It is understandable that if the current frame image is the first frame image captured when behavior recognition processing begins, then there is no previous frame image. Therefore, the initial feature map set can be preset as the first feature map set for behavior recognition processing.

[0095] S204, input the current frame image and the first feature map set into the recognition network model for behavior recognition processing, so as to obtain the target behavior score of the preset behavior corresponding to the current frame image;

[0096] In one embodiment, after obtaining the current frame image and the first feature map set, the current frame image and the first feature map set are output into the recognition network model. The recognition network model is used to perform behavior recognition processing based on the current frame image and the first feature map set to obtain the target behavior score corresponding to the current frame image.

[0097] Specifically, the method for obtaining the target behavior score corresponding to the current frame image can be as follows: the first layer of the convolutional layer in the recognition network model is used to extract features from the current frame image to obtain an initial feature map. The initial feature map is then input into the second layer of the convolutional layer for behavior recognition processing. This process is repeated until the target layer in the convolutional layer is reached. When the target layer performs behavior recognition processing, in addition to using the feature map input from the previous layer of the target layer, the feature map corresponding to the target layer in the first feature map set is also obtained and fused to obtain the feature map corresponding to the target layer. The feature map of the 100th layer of the convolutional layer in the recognition network model is then obtained. Based on the pooling layer and fully connected layer in the recognition network model, the target behavior score corresponding to the preset behavior of the current frame image is calculated.

[0098] For example, if the recognition network model has 100 convolutional layers, and the target layer is a multiple of ten in the convolutional layers, then when the target layer is the tenth layer, the feature map corresponding to the tenth layer is obtained by fusing the feature map of the ninth layer and the first feature map corresponding to the tenth layer in the first feature map set. Figure 4 As shown, Figure 4 When calculating the tenth layer of the convolutional layer in the recognition network model, the feature map of the ninth layer and the first feature map of the tenth layer in the first feature map set are obtained and fused for calculation.

[0099] The target behavior score is the score obtained by the recognition network model for behavior recognition processing. It is used to indicate the probability that a preset behavior exists in the current frame image and the images before the current frame image. The higher the score value, the greater the probability that a preset behavior exists in the current frame image and the images before the current frame image.

[0100] S205, Based on the detection network model, human body parts are detected in the current frame image to determine the region image containing the target part in the current frame image;

[0101] In one embodiment, after acquiring the current frame image, in order to determine whether there is a human body part corresponding to a preset behavior in the current frame image, the current frame image is input into the detection network model, and the human body part is detected in the current frame image based on the detection network model to determine the region image containing the target part in the current frame image.

[0102] The detection network model can be a network model with object detection capabilities, used to perform part detection on the current frame image.

[0103] The target body part can be a body part corresponding to a preset behavior, such as the hand.

[0104] A region image can be an image that includes the target region in the current frame, such as... Figure 5 As shown, Figure 5 The "region image" in the image refers to the image that includes the target part, the "hand".

[0105] S206, Input the region image into the recognition network model for part pose recognition processing to obtain the part pose recognition result;

[0106] In one embodiment, after acquiring the region image, the region image is input into the recognition network model, and the pose of the parts in the region image is recognized according to the recognition network model to obtain the pose recognition result.

[0107] The pose of a part can be the pose adopted by the target part in a preset behavior. For example, if the target part is the hand, the pose recognition result can be a single finger, etc.

[0108] S207, if the part pose recognition result indicates that the target part is in a preset pose, then the first memory value corresponding to the current frame image is set to the initial value;

[0109] In one embodiment, after obtaining the part pose, the part pose recognition result is matched with a preset pose. If the part pose recognition result indicates that the target part is in the preset pose, the first memory value corresponding to the current frame image is set to the initial value.

[0110] The first memory value can be the memory value corresponding to the current frame image, used to record whether the preset pose exists in the current frame image.

[0111] The preset posture can be a pre-defined posture for a part used to activate vehicle functions, such as "drawing a circle with one finger". It should be noted that if the preset state is "drawing a circle with one finger", it can be further limited to one hand performing the "drawing a circle with one finger" operation. If two hands are used to perform "drawing a circle with one finger", the part posture recognition result indicates that the target part does not belong to the preset posture. Specific settings can be configured according to actual needs.

[0112] For example, such as Figure 5 As shown, Figure 5 If the target part of the central region image is the hand, and the pose of the part is a single finger, then the first memory value corresponding to the current frame image is set as the initial value.

[0113] S208, if the part pose recognition result indicates that the target part is in another pose, then the second memory value corresponding to the previous frame image is obtained, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image.

[0114] In one embodiment, after obtaining the part pose, the part pose recognition result is matched with a preset pose. If the part pose recognition result indicates that the target part is in another pose, the second memory value corresponding to the previous frame image is obtained, the second memory value is decremented according to preset data, and the value obtained after decrementing is used as the first memory value corresponding to the current frame image.

[0115] Other poses can be any pose performed by the target part other than the preset pose. For example, if the preset pose is "single finger", then other poses can be "two fingers", etc. It should be noted that if the preset pose is "single finger" and the target part is "one hand", then the current frame image includes two hands, and each hand is "single finger", which is also considered not to conform to the preset pose.

[0116] The second memory value can be the memory value corresponding to the previous frame image. The preset data can be a pre-set value that has been decremented, such as 1.

[0117] For example, if the second memory value is 20 and the preset data is 1, then the first memory value corresponding to the current frame image is subtracted to "20-1", resulting in the first memory value "19" corresponding to the current frame image.

[0118] It should be noted that, since there is a case where the second memory value is 0, in order to avoid the second memory value occupying too much memory due to continuous data decrementing, a feasible method is to assign the first memory value to 0 when decrementing the second memory value if the second memory value is 0.

[0119] S209, if the first memory value is greater than the preset memory value and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior;

[0120] In one embodiment, after obtaining the first memory value and the target behavior score corresponding to the current frame image, the first memory value is compared with a preset memory value, and the target behavior score is compared with a preset score threshold. If the first memory value is greater than the preset memory value and the target behavior score is greater than the preset score threshold, then the behavior recognition result is determined to be the preset behavior.

[0121] The preset score threshold can be used to determine the behavior recognition result corresponding to the current frame image. The target behavior score is compared with the preset score threshold to determine the behavior recognition result corresponding to the current frame image.

[0122] The behavior recognition result can be determined by the target behavior score output by the recognition network model, which corresponds to the recognition result of the current frame image. Specifically, it can include preset behaviors and other behaviors. Among them, other behaviors can be behaviors other than the preset behaviors. For example, if the preset behavior is "drawing a circle with one finger", then other behaviors can be "drawing a line with one finger", "drawing a circle with two fingers", etc.

[0123] For example, if the number of consecutive image frames is 36, the preset number of frames is 24, the first memory value is 24, the preset memory value is 0, the target behavior score is 0.9, and the preset score threshold is 0.8, then since the number of consecutive image frames (36 frames) is greater than the preset number of frames (24 frames), the first memory value (24) is greater than the preset memory value (0), and the target behavior score (0.9) is greater than the preset score threshold (0.8), the behavior recognition result can be determined to be the preset behavior.

[0124] Furthermore, if the target behavior score is less than a preset score threshold, the behavior recognition result corresponding to the current frame image is determined to be another behavior. It can be understood that the target behavior score is the output of the recognition network model for behavior recognition of the current frame image. The higher the target behavior score, the greater the probability that the behavior in the current frame image is a preset behavior. Conversely, if the target behavior score is less than the preset score threshold, it can be considered that the probability of the behavior in the current frame image being a preset behavior is too small, and therefore the behavior is identified as another behavior.

[0125] Furthermore, if the first memory value is less than or equal to the preset memory value, it can be assumed that the target part of the preset posture is not included within the set number of frames. Therefore, even if the target behavior score is greater than the preset score threshold, the behavior recognition result is considered to be other behaviors.

[0126] The set frame number can be a pre-set number of frames used to determine the execution of a preset behavior. For example, if 12 frames are captured per second, the set frame number could be 24 frames. If the first memory value is less than or equal to the preset memory value, it is considered that the user has not performed the preset behavior by adopting a preset posture through the target body part within the set frame number. Therefore, even if the target behavior score obtained by the recognition network model for the current frame image is greater than the preset score threshold, it is considered that a similar behavior to the preset behavior was performed by adopting other postures or other body parts, and the behavior recognition result is determined as other behavior.

[0127] S210, if the number of consecutive image frames for the behavior recognition process is greater than a preset number of frames, and the behavior recognition result is a preset behavior, then control the vehicle to start the vehicle function corresponding to the preset behavior.

[0128] In one embodiment, a comparison result of the number of consecutive image frames for behavior recognition processing and a preset number of frames is obtained. If the number of consecutive image frames is greater than or equal to the preset number of frames, and the behavior recognition result of the current frame image is determined to be a preset behavior, then the vehicle is controlled to start the vehicle function corresponding to the preset behavior.

[0129] The consecutive image frame count can be the number of frames that have undergone behavior recognition processing since the vehicle met the parking scenario conditions. Specifically, it can be counted from the image where behavior recognition processing began. If the vehicle's current scenario state changes from a parking scenario to another scenario, the consecutive image frame count is reset to zero.

[0130] The preset frame rate can be a pre-set number of frames sufficient for feature accumulation in the initial stage of behavior recognition processing. For example, it could be 24 frames.

[0131] Furthermore, if the number of consecutive image frames for behavior recognition processing is less than the preset number of frames, the behavior recognition result is discarded, and the process proceeds to step S203 to perform behavior recognition processing on the next frame of the current frame image.

[0132] It should be noted that the first feature map set obtained when performing behavior recognition processing on the first frame image is the initial feature map set. The feature values ​​of each feature map in the initial feature map set are preset values. In order to improve the accuracy of behavior recognition results, feature accumulation is required. Therefore, if the number of consecutive image frames is greater than or equal to the preset number of frames, it can be considered that the feature accumulation is sufficient to ensure the accuracy of behavior recognition. If the number of consecutive image frames is less than the preset number of frames, in order to avoid recognition errors that may lead to the accidental activation of the vehicle function, the behavior recognition result is marked as an invalid result and discarded.

[0133] For example, a method for controlling the vehicle's activation function can be as follows: if the preset behavior is to draw a circle with one finger, and the vehicle function is to activate the vehicle's relaxation mode, then when it is determined that the driver is drawing a circle with one finger in front of the camera capturing the current frame image, the relevant controllers in the vehicle, such as the window and lighting controllers, control the corresponding components to activate the relaxation mode. In the relaxation mode, all vehicle windows and the sunroof are closed, preset audio data is played at a preset volume, and the vehicle's lights are controlled to illuminate at a preset brightness.

[0134] Understandably, since the activation and maintenance of vehicle functions depend on the matching of the vehicle's current scene state with the parking scene, if the user performs operations such as shifting gears that change the vehicle's current scene state, the vehicle function will be deactivated.

[0135] In this embodiment, a feature map set is obtained by progressively identifying multiple frames of images. A recognition network model is then used to perform behavior recognition on the current frame image and its corresponding feature map set, yielding a behavior recognition result. The recognition network model identifies the feature map set, thus avoiding recognition errors due to timing issues and reducing data processing complexity and efficiency. This improves the convenience and accuracy of vehicle behavior recognition, as well as the ease of user operation in activating vehicle functions. Furthermore, the current scene state of the vehicle is matched with the parking scene state to determine if the current scene state meets the conditions for activating vehicle functions. This prevents accidental activation of vehicle functions even when the scene state is met, thus improving vehicle safety and reliability. Finally, by acquiring and comparing memory values, the pose of the corresponding part in the current frame image is ensured to be a preset pose, thereby improving the accuracy of behavior recognition.

[0136] based on Figure 1 The system architecture will be discussed below. Figure 6 and Figure 7 This application provides a detailed description of the behavior recognition device provided in its embodiments. It should be noted that... Figure 6 The behavior recognition device 1 in the application is used to perform the behavior recognition of this application. Figures 2-5 The methods shown in the embodiments are for illustrative purposes only, illustrating the parts relevant to the embodiments of this application. For specific technical details not disclosed, please refer to this application. Figures 2-5 The example shown.

[0137] The data acquisition unit 11 is used to acquire the current frame image for behavior recognition and to acquire the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion.

[0138] The behavior recognition unit 12 inputs the current frame image and the first feature map set into the recognition network model for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image;

[0139] The function activation unit 13 is used to control the vehicle to activate the vehicle function corresponding to the preset behavior if the number of consecutive image frames for the behavior recognition processing is greater than a preset number of frames and the behavior recognition result is a preset behavior.

[0140] Optionally, the behavior recognition unit 12 is also used for:

[0141] The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the target behavior score of the preset behavior corresponding to the current frame image;

[0142] If the target behavior score is greater than a preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0143] Optionally, the behavior recognition unit 12 is also used for:

[0144] If the target behavior score is less than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be other behavior, which is behavior other than the preset behavior.

[0145] Optional, such as Figure 7 As shown, the behavior recognition device 1 also includes a memory value acquisition unit 14, used for:

[0146] Based on the detection network model, human body parts are detected in the current frame image to determine the region image in the current frame image that contains the target body parts;

[0147] The image of the region is input into the recognition network model for part pose recognition processing to obtain the part pose recognition result;

[0148] If the position recognition result indicates that the target position is in a preset position, then the first memory value corresponding to the current frame image is set to the initial value;

[0149] If the position recognition result indicates that the target position is in another position, then the second memory value corresponding to the previous frame image is obtained, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image.

[0150] The behavior recognition unit 12 is also used for:

[0151] If the first memory value is greater than the preset memory value, and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0152] Optionally, the behavior recognition unit 12 is also used for:

[0153] If the number of consecutive image frames for the behavior recognition process is less than a preset number of frames, the behavior recognition result is discarded, and the process proceeds to the step of obtaining the current frame image for behavior recognition.

[0154] Optionally, the data acquisition unit 11 is also used for:

[0155] If the current frame image is the first frame image, then an initial feature map set is obtained, and the feature value of each initial feature map in the initial feature map set is a preset feature value.

[0156] Optional, such as Figure 7 As shown, the behavior recognition device 1 also includes an image acquisition unit 15, used for:

[0157] Obtain the current scene state of the vehicle and match the current scene state with the parking scene state;

[0158] If the current scene state matches the parking scene state, then control the vehicle's camera to capture the current frame image.

[0159] In this embodiment, a feature map set is obtained by progressively identifying multiple frames of images. A recognition network model is then used to perform behavior recognition on the current frame image and its corresponding feature map set, yielding a behavior recognition result. The recognition network model identifies the feature map set, thus avoiding recognition errors due to timing issues and reducing data processing complexity and efficiency. This improves the convenience and accuracy of vehicle behavior recognition, as well as the ease of user operation in activating vehicle functions. Furthermore, the current scene state of the vehicle is matched with the parking scene state to determine if the current scene state meets the conditions for activating vehicle functions. This prevents accidental activation of vehicle functions even when the scene state is met, thus improving vehicle safety and reliability. Finally, by acquiring and comparing memory values, the pose of the corresponding part in the current frame image is ensured to be a preset pose, thereby improving the accuracy of behavior recognition.

[0160] Please see Figure 8 This provides a structural schematic diagram of a vehicle according to an embodiment of this application. Figure 8 As shown, the vehicle 500 includes a processor 501 and a memory 502. The processor 501 and the memory 502 are electrically connected.

[0161] The processor 501 is the control center of the vehicle 500 and may include one or more processing cores. The processor 501 connects to various parts of the vehicle 500 via various interfaces and lines. It executes various functions and processes data of the vehicle 500 by running or calling computer programs stored in the memory 502 and by calling data stored in the memory 502, thereby providing overall control of the vehicle 500. Optionally, the processor 501 may be implemented using at least one of the following hardware forms: Digital Signal Processing (DSP), Field Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 501 may integrate one or more of the following: CPU, Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user page, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem handles wireless communication. It is understood that the modem may also be implemented separately as a communication chip, without being integrated into the processor 501.

[0162] The memory 502 can be used to store software programs and modules. The processor 501 executes various functional applications and data processing by running the computer programs and modules stored in the memory 502. The memory 502 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, computer programs required for at least one function, etc.; the data storage area may store data created based on the use of the vehicle 500, etc.

[0163] Furthermore, memory 502 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, memory 502 may also include a memory controller to provide processor 501 with access to memory 502.

[0164] In this embodiment, the processor 501 in the vehicle 500 loads the instructions corresponding to the processes of one or more computer programs into the memory 502 according to the following steps, and the processor 501 runs the computer programs stored in the memory 502 to realize various functions, as follows:

[0165] Obtain the current frame image for behavior recognition, and obtain the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion.

[0166] The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the behavior recognition result corresponding to the current frame image;

[0167] If the number of consecutive image frames for the behavior recognition process is greater than a preset number of frames, and the behavior recognition result is a preset behavior, then the vehicle is controlled to activate the vehicle function corresponding to the preset behavior.

[0168] Optionally, when the processor 501 performs behavior recognition processing by inputting the current frame image and the first feature map set into the recognition network model to obtain the behavior recognition result corresponding to the current frame image, it specifically executes:

[0169] The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the target behavior score of the preset behavior corresponding to the current frame image;

[0170] If the target behavior score is greater than a preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0171] Optionally, processor 501 also performs:

[0172] If the target behavior score is less than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be other behavior, which is behavior other than the preset behavior.

[0173] Optionally, when the target motor discharges, after acquiring the current frame image for behavior recognition, the processor 501 further executes:

[0174] Based on the detection network model, human body parts are detected in the current frame image to determine the region image in the current frame image that contains the target body parts;

[0175] The image of the region is input into the recognition network model for part pose recognition processing to obtain the part pose recognition result;

[0176] If the position recognition result indicates that the target position is in a preset position, then the first memory value corresponding to the current frame image is set to the initial value;

[0177] If the position recognition result indicates that the target position is in another position, then the second memory value corresponding to the previous frame image is obtained, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image.

[0178] When processor 501 executes the statement "If the target behavior score is greater than a preset score threshold, then determine that the behavior recognition result corresponding to the current frame image is a preset behavior", it specifically performs the following:

[0179] If the first memory value is greater than the preset memory value, and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior.

[0180] Optionally, when the target motor is charging, after executing the step of inputting the current frame image and the first feature map set into the recognition network model for behavior recognition processing, the processor 501 further executes:

[0181] If the number of consecutive image frames for the behavior recognition process is less than a preset number of frames, the behavior recognition result is discarded, and the process proceeds to the step of obtaining the current frame image for behavior recognition.

[0182] Optionally, when the processor 501 executes the process of obtaining the first feature map set corresponding to the previous frame image of the current frame image, it specifically performs the following:

[0183] If the current frame image is the first frame image, then an initial feature map set is obtained, and the feature value of each initial feature map in the initial feature map set is a preset feature value.

[0184] Optionally, before acquiring the current frame image for action recognition, processor 501 also performs:

[0185] Obtain the current scene state of the vehicle and match the current scene state with the parking scene state;

[0186] If the current scene state matches the parking scene state, then control the vehicle's camera to capture the current frame image.

[0187] In this embodiment, a feature map set is obtained by progressively identifying multiple frames of images. A recognition network model is then used to perform behavior recognition on the current frame image and its corresponding feature map set, yielding a behavior recognition result. The recognition network model identifies the feature map set, thus avoiding recognition errors due to timing issues and reducing data processing complexity and efficiency. This improves the convenience and accuracy of vehicle behavior recognition, as well as the ease of user operation in activating vehicle functions. Furthermore, the current scene state of the vehicle is matched with the parking scene state to determine if the current scene state meets the conditions for activating vehicle functions. This prevents accidental activation of vehicle functions even when the scene state is met, thus improving vehicle safety and reliability. Finally, by acquiring and comparing memory values, the pose of the corresponding part in the current frame image is ensured to be a preset pose, thereby improving the accuracy of behavior recognition.

[0188] Please see Figure 9 This provides a structural schematic diagram of a vehicle according to an embodiment of this application. Figure 9 As shown, the vehicle 500 includes: a processor 501, a memory 502, a display screen 503, a camera assembly 504, an audio circuit 505, a sensor 506, and a power supply 507. The processor 501 is electrically connected to the display screen 503, the camera assembly 504, the audio circuit 505, the sensor 506, and the power supply 507.

[0189] The display screen 503 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces of the vehicle 500, which can be composed of images, text, icons, videos, and any combination thereof.

[0190] The camera assembly 504 may include image processing circuitry, which may be implemented using hardware and / or software components and may include various processing units defining an image signal processing (ISP) pipeline. The image processing circuitry may include at least: multiple cameras, an image signal processor (ISP), control logic, and image memory. Each camera may include at least one or more lenses and an image sensor. The image sensor may include a color filter array (such as a Bayer filter). The image sensor acquires light intensity and wavelength information captured by each imaging pixel of the image sensor and provides a set of raw image data that can be processed by the image signal processor.

[0191] The audio circuit 505 provides an audio interface between the user and the vehicle 500 via a speaker and microphone. The audio circuit 505 includes a microphone, which is electrically connected to the processor 501. The microphone is used to receive voice input from the user.

[0192] Sensor 506 is used to collect information about the vehicle 500 itself, user information, or external environmental information. For example, sensor 506 may include one or more of the following sensors: vibration sensor, temperature sensor, distance sensor, magnetic field sensor, light sensor, acceleration sensor, fingerprint sensor, Hall sensor, position sensor, gyroscope, inertial sensor, attitude sensor, barometer, heart rate sensor, etc.

[0193] Power supply 507 is used to supply power to various components of vehicle 500. In some embodiments, power supply 507 can be logically connected to processor 501 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0194] It should be understood that the apparatus provided in this application embodiment is used to perform the above-described behavior recognition method, and therefore can achieve the same effect as the above-described implementation method.

[0195] When using an integrated unit, the device may include a processing module and a storage module. When the device is applied to a vehicle, the processing module can be used to control and manage the vehicle's movements. The storage module can be used to support the vehicle in executing program code, etc.

[0196] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits as disclosed in this application. The processor may also be a combination of computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and microprocessors, etc., and the storage module may be a memory.

[0197] In addition, the device provided in this application embodiment may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a behavior recognition method provided in the above embodiment.

[0198] This application also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the behavior recognition method provided in the above embodiments.

[0199] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to implement the behavior recognition method provided in the above embodiment.

[0200] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0201] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0202] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0203] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A behavior recognition method, characterized in that, The method includes: Obtain the current frame image for behavior recognition, and obtain the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion. Based on the detection network model, human body parts are detected in the current frame image to determine the region image in the current frame image that contains the target body parts; The image of the region is input into the recognition network model for part pose recognition processing to obtain the part pose recognition result; If the part pose recognition result indicates that the target part is a preset pose, then the first memory value corresponding to the current frame image is set to the initial value. The first memory value is the memory value corresponding to the current frame image, which is used to record whether the preset pose exists in the current frame image. If the position recognition result indicates that the target position is in another position, then the second memory value corresponding to the previous frame image is obtained, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image. The second memory value is the memory value corresponding to the previous frame image. The current frame image and the first feature map set are input into the recognition network model for behavior recognition processing to obtain the target behavior score of the preset behavior corresponding to the current frame image; If the first memory value is greater than the preset memory value, and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior; If the number of consecutive image frames for the behavior recognition process is greater than a preset number of frames, and the behavior recognition result is a preset behavior, then the vehicle is controlled to activate the vehicle function corresponding to the preset behavior.

2. The method according to claim 1, characterized in that, The method further includes: If the target behavior score is less than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be other behavior, which is behavior other than the preset behavior.

3. The method according to claim 1, characterized in that, After the step of inputting the current frame image and the first feature map set into the recognition network model for behavior recognition processing, the method further includes: If the number of consecutive image frames for the behavior recognition process is less than a preset number of frames, the behavior recognition result is discarded, and the process proceeds to the step of obtaining the current frame image for behavior recognition.

4. The method according to claim 1, characterized in that, The step of obtaining the first feature map set corresponding to the previous frame image of the current frame image includes: If the current frame image is the first frame image, then an initial feature map set is obtained, and the feature value of each initial feature map in the initial feature map set is a preset feature value.

5. The method according to claim 1, characterized in that, Before acquiring the current frame image for behavior recognition, the method further includes: Obtain the current scene state of the vehicle and match the current scene state with the parking scene state; If the current scene state matches the parking scene state, then control the vehicle's camera to capture the current frame image.

6. A behavior recognition device, characterized in that, The device includes: The data acquisition unit is used to acquire the current frame image for behavior recognition and to acquire the first feature map set corresponding to the previous frame image of the current frame image. The first feature map set is the feature map set obtained after the previous frame image is input into the convolutional layer of the recognition network model for feature map fusion. The memory value acquisition unit is used to perform human body part detection on the current frame image based on the detection network model, and determine the region image containing the target part in the current frame image; input the region image into the recognition network model for part pose recognition processing to obtain part pose recognition result; if the part pose recognition result indicates that the target part is a preset pose, then the first memory value corresponding to the current frame image is set to an initial value, the first memory value is the memory value corresponding to the current frame image, used to record whether the preset pose exists in the current frame image; if the part pose recognition result indicates that the target part is another pose, then the second memory value corresponding to the previous frame image is acquired, and the second memory value is decremented according to preset data to obtain the first memory value corresponding to the current frame image, the second memory value is the memory value corresponding to the previous frame image; The behavior recognition unit is used to input the current frame image and the first feature map set into the recognition network model for behavior recognition processing, so as to obtain the target behavior score of the preset behavior corresponding to the current frame image; if the first memory value is greater than the preset memory value and the target behavior score is greater than the preset score threshold, then the behavior recognition result corresponding to the current frame image is determined to be the preset behavior. The function activation unit is used to control the vehicle to activate the vehicle function corresponding to the preset behavior if the number of consecutive image frames for the behavior recognition processing is greater than a preset number of frames and the behavior recognition result is a preset behavior.

7. A vehicle, characterized in that, The vehicles include: Memory, used to store executable program code; A processor for calling and running the executable program code from the memory, causing the vehicle to perform the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program code that, when executed, implements the method as described in any one of claims 1 to 5.

Images