Parking safety monitoring method and device based on AI and IoT technologies

Abstract

This invention relates to the field of vehicle safety technology and discloses a parking safety monitoring method and device based on AI and IoT technologies. The method includes: calculating a target judgment value for the vehicle based on multi-dimensional feature information of the detected object detected by a detection device corresponding to the vehicle; determining an adaptive judgment threshold for the vehicle regarding the environment based on environmental information of the vehicle's surroundings; determining whether the vehicle meets preset wake-up conditions based on the judgment value and the adaptive judgment threshold; and controlling the vehicle to enter a wake-up state when the wake-up conditions are met, in order to perform corresponding vehicle safety monitoring operations. Therefore, implementing this invention can improve the accuracy and reliability of vehicle monitoring, thereby reducing the probability of false vehicle wake-up while maintaining good parking safety monitoring performance.

Description

Technical Field

[0001] This invention relates to the field of vehicle safety technology, and in particular to a parking safety monitoring method and device based on AI and IoT technologies. Background Technology

[0002] With the development of intelligent vehicles, parking safety monitoring technology has received increasing attention. Currently, parking safety monitoring methods are mainly based on millimeter-wave radar technology. This technology can monitor the surrounding environment of a vehicle when it is parked and unattended, and can wake up the vehicle to promptly alert the owner or take appropriate measures when abnormalities are detected, thus improving the effectiveness of parking safety monitoring.

[0003] However, in practice, it has been found that existing parking safety monitoring methods simply use the time distance (distance-to-speed ratio) between the vehicle and surrounding objects detected by millimeter-wave radar, along with a preset time distance threshold based on empirical values, to determine the type of object approaching the vehicle. This approach is relatively simplistic and easily leads to a high probability of the vehicle being falsely activated. Therefore, it is crucial to develop a technical solution that maintains good parking safety monitoring performance while reducing the probability of false vehicle activation. Summary of the Invention This invention provides a parking safety monitoring method and device based on AI and IoT technologies, which can reduce the probability of vehicles being falsely awakened while maintaining good parking safety monitoring results.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a parking safety monitoring method based on AI and IoT technologies, the method comprising: Based on the multi-dimensional feature information of the detected object detected by the detection equipment corresponding to the vehicle, the target judgment value corresponding to the vehicle is calculated. By determining the environmental information of the environment in which the vehicle is located, an adaptive judgment threshold for the vehicle with respect to the environment is calculated; Based on the target judgment value and the adaptive judgment threshold, determine whether the vehicle meets the preset wake-up conditions; When it is determined that the vehicle meets the wake-up conditions, the vehicle is controlled to enter the wake-up state, and after the vehicle is woken up, the safety monitoring device matching the detection device is controlled to perform corresponding safety monitoring operations on the vehicle.

[0005] As an optional implementation, in the first aspect of the present invention, the step of calculating the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle includes: Obtain detection information from the detection device corresponding to the vehicle, wherein the detection information includes at least one of the starting frequency bandwidth and the frequency sweep period; Based on the detection information, the multi-dimensional feature information of the detected object is calculated. The detected object is an object within a preset range where the vehicle is located. The multi-dimensional feature information includes multiple sub-class feature information, and all the sub-class feature information includes at least two of the following: speed, distance, micro-Doppler features, and RCS value. For each of the subclass feature information, calculate the judgment value of the subclass feature information and determine the weight coefficient corresponding to the subclass feature information; The target judgment value corresponding to the vehicle is calculated based on the judgment values ​​of all the sub-class feature information and the weight coefficient corresponding to each sub-class feature information.

[0006] As an optional implementation, in the first aspect of the present invention, the calculation of the judgment value of the subclass feature information includes: When the subclass feature information is the micro-Doppler feature, the phase timing signal of the detection device is determined based on the detection information; Perform a short-time Fourier transform on the phase timing signal to obtain a time spectrum; and extract the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time spectrum as the main modulation frequency; If the main modulation frequency is greater than or equal to the first preset frequency threshold and less than or equal to the second preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the first judgment value, where the first preset frequency threshold is less than the second preset frequency threshold. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the second judgment value, and the second preset frequency threshold is less than the third preset frequency threshold. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the third judgment value, where the first judgment value is greater than the second judgment value, and the second judgment value is greater than the third judgment value.

[0007] As an optional implementation, in the first aspect of the present invention, the step of calculating the adaptive judgment threshold of the vehicle with respect to the environment by determining the environmental information of the environment in which the vehicle is located includes: Obtain the vehicle's location information and the vehicle's wake-up record information; Based on the vehicle's location information, the environmental type of the environment in which the vehicle is located is determined, and the environmental type has a corresponding risk level; based on the risk level corresponding to the environmental type, a baseline judgment value corresponding to the environmental type is determined, and the risk level is negatively correlated with the baseline judgment value; Based on the vehicle's wake-up record information, the wake-up record feature value of the vehicle is calculated, and the wake-up record feature value is calculated under the condition that the vehicle is under the same location information; Based on the baseline judgment value corresponding to the environment type and the wake-up record feature value of the vehicle, the adaptive judgment threshold of the vehicle with respect to the environment is calculated.

[0008] As an optional implementation, in the first aspect of the present invention, determining whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold includes: The comparison result is obtained by comparing the target judgment value with the adaptive judgment threshold; When the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it is determined that the vehicle meets the preset wake-up conditions; When the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it is determined that the vehicle does not meet the preset wake-up conditions.

[0009] As an optional implementation, in the first aspect of the present invention, the safety monitoring device matching the detection device performs corresponding safety monitoring operations on the vehicle, including: Based on the location of the detection device, control a matching safety monitoring device to monitor the vehicle-related image data. The image data is analyzed to determine whether the vehicle meets the preset warning conditions; When it is determined that the vehicle meets the warning conditions, a warning notification is sent to the owner of the vehicle.

[0010] As an optional implementation, in the first aspect of the present invention, the image data includes video data and / or image data; And, the analysis of the image data to determine whether the vehicle meets the preset warning conditions includes: When the image data is video data, the video data is split to obtain multiple initial video frames corresponding to the video data; Each initial video frame is input into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame; Based on the object type of the foreground object indicated by all the initial video frames, select all target video frames whose object type is a preset type from all the initial video frames; Calculate the ratio between the number of target video frames and the number of initial video frames, and determine whether the ratio is greater than a preset ratio; When it is determined that the ratio is greater than the preset ratio, the vehicle is determined to meet the preset warning conditions; When it is determined that the ratio is less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions.

[0011] A second aspect of this invention discloses a parking safety monitoring device based on AI and IoT technologies, the device comprising: The calculation module is used to calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle. The calculation module is also used to calculate the adaptive judgment threshold of the vehicle with respect to the environment by determining the environmental information of the environment in which the vehicle is located; The judgment module is used to determine whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold. The control module is used to control the vehicle to enter the wake-up state when the judgment module determines that the vehicle meets the wake-up conditions, and after the vehicle is woken up, control the safety monitoring device matched with the detection device to perform corresponding safety monitoring operations on the vehicle.

[0012] As an optional implementation, in the second aspect of the present invention, the specific method by which the calculation module calculates the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle includes: Obtain detection information from the detection device corresponding to the vehicle, wherein the detection information includes at least one of the starting frequency bandwidth and the frequency sweep period; Based on the detection information, the multi-dimensional feature information of the detected object is calculated. The detected object is an object within a preset range where the vehicle is located. The multi-dimensional feature information includes multiple sub-class feature information, and all the sub-class feature information includes at least two of the following: speed, distance, micro-Doppler features, and RCS value. For each of the subclass feature information, calculate the judgment value of the subclass feature information and determine the weight coefficient corresponding to the subclass feature information; The target judgment value corresponding to the vehicle is calculated based on the judgment values ​​of all the sub-class feature information and the weight coefficient corresponding to each sub-class feature information.

[0013] As an optional implementation, in the second aspect of the present invention, the specific method by which the calculation module calculates the judgment value of the subclass feature information includes: When the subclass feature information is the micro-Doppler feature, the phase timing signal of the detection device is determined based on the detection information; Perform a short-time Fourier transform on the phase timing signal to obtain a time spectrum; and extract the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time spectrum as the main modulation frequency; If the main modulation frequency is greater than or equal to the first preset frequency threshold and less than or equal to the second preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the first judgment value, where the first preset frequency threshold is less than the second preset frequency threshold. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the second judgment value, and the second preset frequency threshold is less than the third preset frequency threshold. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the third judgment value, where the first judgment value is greater than the second judgment value, and the second judgment value is greater than the third judgment value.

[0014] As an optional implementation, in a second aspect of the present invention, the specific method by which the calculation module calculates the adaptive judgment threshold of the vehicle with respect to the environment by determining the environmental information of the environment in which the vehicle is located includes: Obtain the vehicle's location information and the vehicle's wake-up record information; Based on the vehicle's location information, the environmental type of the environment in which the vehicle is located is determined, and the environmental type has a corresponding risk level; based on the risk level corresponding to the environmental type, a baseline judgment value corresponding to the environmental type is determined, and the risk level is negatively correlated with the baseline judgment value; Based on the vehicle's wake-up record information, the wake-up record feature value of the vehicle is calculated, and the wake-up record feature value is calculated under the condition that the vehicle is under the same location information; Based on the baseline judgment value corresponding to the environment type and the wake-up record feature value of the vehicle, the adaptive judgment threshold of the vehicle with respect to the environment is calculated.

[0015] As an optional implementation, in a second aspect of the present invention, the specific method by which the judging module determines whether the vehicle meets the preset wake-up conditions based on the target judging value and the adaptive judging threshold includes: The comparison result is obtained by comparing the target judgment value with the adaptive judgment threshold; When the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it is determined that the vehicle meets the preset wake-up conditions; When the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it is determined that the vehicle does not meet the preset wake-up conditions.

[0016] As an optional implementation, in a second aspect of the invention, the control module controls a safety monitoring device matched with the detection device to perform corresponding safety monitoring operations on the vehicle, including: Based on the location of the detection device, control a matching safety monitoring device to monitor the vehicle-related image data. The image data is analyzed to determine whether the vehicle meets the preset warning conditions; When it is determined that the vehicle meets the warning conditions, a warning notification is sent to the owner of the vehicle.

[0017] As an optional implementation, in a second aspect of the invention, the image data includes video data and / or image data; Furthermore, the control module analyzes the image data to determine whether the vehicle meets preset warning conditions, including: When the image data is video data, the video data is split to obtain multiple initial video frames corresponding to the video data; Each initial video frame is input into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame; Based on the object type of the foreground object indicated by all the initial video frames, select all target video frames whose object type is a preset type from all the initial video frames; Calculate the ratio between the number of target video frames and the number of initial video frames, and determine whether the ratio is greater than a preset ratio; When it is determined that the ratio is greater than the preset ratio, the vehicle is determined to meet the preset warning conditions; When it is determined that the ratio is less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions.

[0018] A third aspect of the present invention discloses another parking safety monitoring device based on AI and IoT technologies, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the parking safety monitoring method based on AI and IoT technologies disclosed in the first aspect of the present invention.

[0019] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute the parking safety monitoring method based on AI and IoT technologies disclosed in the first aspect of the present invention.

[0020] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: In this embodiment of the invention, a target judgment value corresponding to the vehicle is calculated based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle; an adaptive judgment threshold for the vehicle regarding the environment is determined based on the environmental information of the vehicle's environment; the vehicle is judged to meet the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold; when the vehicle is judged to meet the wake-up conditions, the vehicle is controlled to enter the wake-up state, and after the vehicle is woken up, the safety monitoring device matching the detection device is controlled to perform corresponding safety monitoring operations on the vehicle. As can be seen, implementing this invention can accurately calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object, and automatically determine the adaptive judgment threshold of the vehicle with respect to the environment based on the environmental information of the vehicle's environment. Then, based on the target judgment value and the adaptive judgment threshold, it can accurately determine whether the vehicle meets the preset wake-up conditions. This allows the judgment threshold used as the basis for vehicle wake-up judgment to adaptively change with the changes in the vehicle's environment, making it more applicable to different scenarios and reducing the probability of false judgment, thereby reducing the probability of the vehicle being falsely woken up. Moreover, after the vehicle is woken up, the safety monitoring equipment matched with the detection equipment can perform corresponding safety monitoring operations on the vehicle, which can improve the matching degree between the detection equipment and the safety monitoring equipment with control requirements, thereby improving the accuracy and efficiency of controlling the safety monitoring equipment and maintaining a good safety monitoring effect. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 2This is a flowchart illustrating another parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 3 This is a flowchart illustrating another parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a parking safety monitoring device disclosed in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a parking safety monitoring device based on AI and IoT technologies disclosed in an embodiment of the present invention; Figure 6 This is a schematic diagram of another parking safety monitoring device based on AI and IoT technologies disclosed in an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0025] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0026] This invention discloses a parking safety monitoring method and device based on AI and IoT technologies. It can accurately calculate the target judgment value corresponding to a vehicle based on the multi-dimensional feature information of the detected object, and automatically determine the vehicle's adaptive judgment threshold based on the environmental information of the vehicle's environment. Then, based on the target judgment value and the adaptive judgment threshold, it accurately determines whether the vehicle meets the preset wake-up conditions. This allows the judgment threshold used as the basis for vehicle wake-up judgment to adaptively change with the vehicle's environment, making it better applicable to different scenarios and reducing the probability of false judgments, thus reducing the probability of the vehicle being falsely woken up. Furthermore, after the vehicle is woken up, the matching safety monitoring equipment controlled by the detection device performs corresponding safety monitoring operations on the vehicle, improving the matching degree between the detection device and the safety monitoring equipment with control requirements, thereby improving the accuracy and efficiency of controlling the safety monitoring equipment and maintaining a good safety monitoring effect. Detailed descriptions follow.

[0027] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 1 The described parking safety monitoring method based on AI and IoT technologies can be applied to parking safety monitoring devices based on AI and IoT technologies. These devices may include monitoring equipment or a monitoring server, which may include a cloud server or a local server; this embodiment of the invention does not impose limitations. Figure 1 As shown, this parking safety monitoring method based on AI and IoT technologies may include the following operations: 101. Based on the multi-dimensional feature information of the detected object detected by the detection equipment corresponding to the vehicle, calculate the target judgment value corresponding to the vehicle.

[0028] In this embodiment of the invention, the detection device may be a millimeter-wave radar, a lidar, or other devices that can perform the same detection function. This embodiment of the invention does not limit the type of device.

[0029] In this embodiment of the invention, the object being detected is an object within a preset range where the vehicle is located, such as an object detected by the detection device within a range of 5 meters from the vehicle.

[0030] 102. By determining the environmental information of the vehicle's environment, calculate the vehicle's adaptive judgment threshold regarding the environment.

[0031] In this embodiment of the invention, the environmental information may optionally include the environmental type, or it may include at least one of the following: pedestrian and vehicle information in the scene, such as the number of pedestrians, pedestrian density, number of vehicles, vehicle density, etc. This embodiment of the invention does not limit the information.

[0032] In specific embodiments of the present invention, the environmental type of the vehicle's environment is determined, and an adaptive judgment threshold for the vehicle regarding the environment is calculated based on the environmental type; or, pedestrian and vehicle-related information in the scene where the vehicle is located is determined, and an adaptive judgment threshold for the vehicle regarding the environment is calculated based on the pedestrian and vehicle-related information.

[0033] In this embodiment of the invention, there is no order between step 102 and step 101. That is, step 102 can occur before step 101, after step 101, or simultaneously with step 101. This embodiment of the invention does not impose any limitations.

[0034] 103. Based on the target judgment value and the adaptive judgment threshold, determine whether the vehicle meets the preset wake-up conditions.

[0035] In this embodiment of the invention, optionally, the wake-up condition can be used to represent a condition where the target judgment value is greater than the adaptive judgment threshold, or it can be used to represent a condition where the absolute difference between the target judgment value and the adaptive judgment threshold is greater than a preset difference. This embodiment of the invention does not limit the condition.

[0036] In this embodiment of the invention, when the judgment result of step 103 is yes, that is, when it is determined that the vehicle meets the wake-up condition, step 104 is triggered; when the judgment result of step 103 is no, that is, when it is determined that the vehicle does not meet the wake-up condition, the operation of steps 101-102 can be re-executed, or the process can be terminated. This embodiment of the invention does not impose any limitations.

[0037] 104. Control the vehicle to enter the wake-up state, and after the vehicle is woken up, control the safety monitoring equipment that matches the detection equipment to perform corresponding safety monitoring operations on the vehicle.

[0038] In this embodiment of the invention, the safety monitoring device may optionally be a camera or a sensor, such as an infrared sensor; this embodiment of the invention is not limited thereto. For example, after the vehicle is activated, the camera matching the millimeter-wave radar is controlled by the vehicle's imaging device to perform corresponding safety monitoring operations on the vehicle.

[0039] It is evident that implementation Figure 1The described parking safety monitoring method based on AI and IoT technologies can accurately calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object, and automatically determine the adaptive judgment threshold of the vehicle with respect to the environment based on the environmental information of the vehicle's location. Then, based on the target judgment value and the adaptive judgment threshold, it can accurately determine whether the vehicle meets the preset wake-up conditions. This allows the judgment threshold used as the basis for vehicle wake-up judgment to adaptively change with the changes in the vehicle's environment, making it more applicable to different scenarios and reducing the probability of false judgment, thereby reducing the probability of the vehicle being falsely woken up. Furthermore, after the vehicle is woken up, the matching safety monitoring equipment controlled by the detection equipment can perform corresponding safety monitoring operations on the vehicle, which can improve the matching degree between the detection equipment and the safety monitoring equipment with control requirements, thereby improving the accuracy and efficiency of controlling the safety monitoring equipment and maintaining a good safety monitoring effect.

[0040] In an optional embodiment, step 101 above, which calculates the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle, may include: Obtain detection information from the vehicle's corresponding detection equipment; Based on the detection information, calculate the multi-dimensional feature information of the detected object detected by the detection device. The multi-dimensional feature information includes multiple sub-category feature information. For each subclass feature information, calculate the judgment value of the subclass feature information and determine the weight coefficient corresponding to the subclass feature information; The target judgment value corresponding to the vehicle is calculated based on the judgment values ​​of all subclass feature information and the weight coefficients corresponding to each subclass feature information.

[0041] In this embodiment of the invention, optionally, the detection information includes at least one of the starting frequency bandwidth and the sweep frequency period, but this embodiment of the invention does not limit it.

[0042] In this embodiment of the invention, optionally, all sub-class feature information of the multi-dimensional feature information includes at least two of the following: speed, distance, micro-Doppler features, and RCS value (reflection cross-section). Micro-Doppler features can be used to represent the volume of the detected object, and RCS values ​​can be used to represent the type of the detected object (e.g., pedestrian type, or small animal type). This embodiment of the invention does not impose any limitations. This approach considers not only the speed and distance of objects approaching the vehicle, but also the judgment values ​​of RCS and micro-Doppler features. In addition to increasing the probability of waking the vehicle when the object's speed is faster and the distance is closer, it also judges the object's type and volume. Therefore, if the object is human, it can be prioritized for waking. Furthermore, a larger volume also increases the probability of waking. This reduces the probability of false waking caused by small animals approaching the vehicle.

[0043] In this embodiment of the invention, the formula for calculating the distance to the detected object is as follows: ; in, For distance, The peak position of the spectrum. At the speed of light, For the scan cycle, This represents the scan bandwidth.

[0044] In this embodiment of the invention, the formula for calculating the velocity of the detected object is as follows: ; in, Indicates speed, Indicates Doppler frequency shift, Indicates wavelength.

[0045] In this embodiment of the invention, the formula for calculating the RCS value of the detected object is as follows: RCS=(P_r·(4π)^3·d^4) / (P_t·G²·λ²); Where P_r is the target echo power, P_t is the radar transmit power, G is the antenna gain, and λ represents the wavelength.

[0046] In this embodiment of the invention, the micro-Doppler feature of the detected object can be a micro-Doppler frequency threshold, which can be used to distinguish between pedestrians and vehicles. Specifically, when the micro-Doppler frequency threshold is between 1 and 5 Hz, it represents the frequency of pedestrian limb swaying; when the micro-Doppler frequency threshold is between 5 and 50 Hz, it represents the frequency of vehicle vibration.

[0047] In this embodiment of the invention, specifically, for each sub-category of feature information among distance, speed, and RCS value, the sub-category feature information is normalized accordingly to obtain the judgment value of the sub-category feature information.

[0048] In this embodiment of the invention, the formula for calculating the target judgment value corresponding to the vehicle is as follows: ; in, For the judgment value, , , , There are 4 weighting coefficients. This indicates that d is normalized, and vs is a preset speed threshold. Indicates to Normalization is performed. Indicates to Normalization is performed. This indicates the preset comparison value. The judgment value is the micro-Doppler feature.

[0049] In this embodiment of the invention, the value of each weight coefficient can be determined based on multi-scenario experimental verification. Specifically, 200 sets of data (including targets such as pedestrians, vehicles, and small animals) were collected from each of the roadside, underground parking garage, and outdoor parking lot. The false wake-up rate under different weight combinations was statistically analyzed, and the statistical results of the false wake-up rate are shown in Table 1.

[0050] As shown in Table 1, the optimal weighting coefficients are α1=0.3, α2=0.3, α3=0.2, and α4=0.2.

[0051] The preset speed threshold is shown in the following example: vs=10m / s. This speed threshold can be set as follows: According to traffic research data, pedestrian walking speed is usually 0.5–2m / s, and running speed can reach 5–7m / s. Therefore, 10m / s is used to distinguish between normal approach and rapid threat. For example, a speed of less than or equal to 10m / s indicates that the pedestrian is approaching the vehicle at a normal speed, while a speed of more than 10m / s indicates that the pedestrian is approaching the vehicle rapidly and may threaten the vehicle.

[0052] The preset contrast value is shown in the example below: sRCS=300m². This contrast value can be set as follows: the average RCS of small animals (such as cats and dogs) is <1m², the average RCS of humans is between 1 and 2m², and the average RCS of vehicles is >10m². Using 300m² can filter out more than 99% of interference from small animals.

[0053] As can be seen, this optional embodiment can calculate the multi-dimensional feature information (including multiple sub-class feature information) of the detected object detected by the detection device based on the detection information of the vehicle-related detection device, thereby calculating the judgment value of each sub-class feature information and determining the weight coefficient corresponding to that sub-class feature information. Based on the judgment values ​​of all sub-class feature information and the weight coefficient corresponding to each sub-class feature information, the target judgment value corresponding to the vehicle can be accurately calculated, which can improve the accuracy and reliability of the calculation of the target judgment value corresponding to the vehicle. This is beneficial to improving the accuracy of the judgment of whether the vehicle meets the wake-up conditions based on the target judgment value, and thus helps to reduce the probability of the vehicle being falsely woken up.

[0054] In this optional embodiment, as an optional implementation method, calculating the judgment value of the subclass feature information may include: When the subclass feature information is a micro-Doppler feature, the phase timing signal of the detection device is determined based on the detection information; Perform a short-time Fourier transform on the phase-time signal to obtain the time spectrum; and extract the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time spectrum as the main modulation frequency; If the main modulation frequency is greater than or equal to the first preset frequency threshold and less than or equal to the second preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the first judgment value, and the first preset frequency threshold is less than the second preset frequency threshold. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the second judgment value, and the second preset frequency threshold is less than the third preset frequency threshold. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the third judgment value, the first judgment value is greater than the second judgment value, and the second judgment value is greater than the third judgment value.

[0055] For example, if the main modulation frequency is greater than or equal to a first preset frequency threshold and less than or equal to a second preset frequency threshold, the judgment value of the micro-Doppler feature is determined to be 1. In this case, it can be further deduced that the RCS value represents a higher probability of a pedestrian. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to a third preset frequency threshold, the judgment value of the micro-Doppler feature is determined to be 0.5. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, the judgment value of the micro-Doppler feature is determined to be 0. The first preset frequency threshold can be 1 Hz, the second preset frequency threshold can be 5 Hz, and the third preset frequency threshold can be 50 Hz.

[0056] As can be seen, this optional implementation can automatically determine the phase timing signal of the detection device based on the detection information for micro-Doppler features, and perform a short-time Fourier transform on the phase timing signal to obtain a time-frequency spectrum. It then extracts the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time-frequency spectrum as the main modulation frequency. This improves the accuracy of extracting the main modulation frequency used as the basis for determining the judgment value of the micro-Doppler feature. If the main modulation frequency is greater than or equal to a first preset frequency threshold and less than or equal to a second preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the first judgment value. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to a third preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the second judgment value. If the main modulation frequency is less than the first preset frequency threshold or greater than the third preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the third judgment value. By comparing the results of multiple preset frequency thresholds with the main modulation frequency, the accuracy and reliability of determining the judgment value of the micro-Doppler feature can be improved, thereby enhancing the accuracy and reliability of subsequently calculating the target judgment value corresponding to the vehicle based on the judgment value of the micro-Doppler feature.

[0057] In another optional embodiment, step 102 above, which involves determining the environmental information of the vehicle's environment and calculating the vehicle's adaptive judgment threshold regarding the environment, may include: Obtain the vehicle's location information and wake-up record information; Based on the vehicle's location information, determine the environmental type of the vehicle's location, and the corresponding risk level of the environmental type. Based on the risk level corresponding to the environmental type, determine the benchmark judgment value corresponding to the environmental type; Based on the vehicle's wake-up record information, the wake-up record feature value of the vehicle is calculated. The wake-up record feature value is calculated under the condition that the vehicle is in the same location information. Based on the baseline judgment value corresponding to the environment type and the vehicle's wake-up record feature value, calculate the vehicle's adaptive judgment threshold regarding the environment.

[0058] In this embodiment of the invention, optionally, the environment type can be set to at least one of roadside type, underground parking garage type, and outdoor parking lot type, wherein the underground parking garage type can include the company's parking garage and the parking garage where the user lives. Roadside type, outdoor parking lot type, and underground parking garage type correspond to different risk levels, and this embodiment of the invention does not limit them.

[0059] In this embodiment of the invention, the risk level is negatively correlated with the benchmark judgment value; that is, the higher the risk level of an environment type, the smaller its corresponding benchmark judgment value. For example, the benchmark judgment values ​​for roadside type, outdoor parking lot type, and underground garage type can be 0.5, 0.75, and 1, respectively, and this embodiment of the invention does not impose any limitations.

[0060] In this embodiment of the invention, the vehicle's wake-up record information can be used to calculate the vehicle's false alarm rate. The higher the false alarm rate, the larger the corresponding wake-up record feature value is set. Specifically, the calculation formula for the wake-up record feature value is as follows: ; in, To wake up the recorded feature value, Under the condition of the same location, the nearest In this series of wake-ups, the number of false wake-ups refers to the situation where the content captured after being woken up and taking pictures does not meet the warning conditions. For example, the settings in this invention... The value is 20. If the cumulative number of wake-ups is less than 20, then... The value is set to the cumulative number of wake-ups.

[0061] In this embodiment of the invention, the specific formula for calculating the adaptive judgment threshold can be as follows: ; in, To adaptively determine the threshold, The characteristic value of the vehicle's environmental type. As the benchmark judgment value, The weight (e.g., it can be set to 0.5) can be adjusted according to actual needs.

[0062] In this embodiment of the invention, optionally, the baseline judgment value can be determined by setting different values ​​for the vehicle in different environmental types and conducting a limited number of trials. The value with the lowest false wake-up probability is then used as the adaptive judgment threshold. That is, under the same test conditions, different values ​​are set as the baseline judgment value to statistically analyze the probability of false wake-up. This makes the adaptive judgment threshold no longer a fixed value, but rather adaptable to the vehicle's environmental type and wake-up records. This makes it more applicable to different scenarios and effectively reduces the probability of false wake-up. An example experiment is as follows: Ten vehicles are deployed in each of the three scenarios, and the number of false wake-ups is recorded over seven days, as shown in Table 2.

[0063] As shown in Table 2, the adaptive judgment threshold can significantly reduce the false wake-up rate in high-interference scenarios.

[0064] As can be seen, this optional embodiment can automatically determine the environment type of the vehicle's environment based on the acquired vehicle positioning information, and determine the benchmark judgment value corresponding to the environment type based on the risk level corresponding to the environment type. It can improve the accuracy and efficiency of determining the benchmark judgment value based on the risk level corresponding to the environment type, and can quickly and accurately calculate the vehicle's wake-up record feature value based on the vehicle's wake-up record information. Thus, based on the benchmark judgment value and the wake-up record feature value, it can calculate the vehicle's adaptive judgment threshold with respect to the environment, thereby improving the accuracy and reliability of calculating the adaptive judgment threshold.

[0065] In another optional embodiment, step 103 above, determining whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold, may include: The comparison results are obtained by comparing the target judgment value with the adaptive judgment threshold; When the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it is determined that the vehicle meets the preset wake-up conditions; When the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it is determined that the vehicle does not meet the preset wake-up conditions.

[0066] As can be seen, this optional embodiment can compare the target judgment value and the adaptive judgment threshold, and when the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it determines that the vehicle meets the preset wake-up conditions, thereby improving the accuracy and reliability of determining that the vehicle meets the preset wake-up conditions. Conversely, when the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it determines that the vehicle does not meet the preset wake-up conditions, thereby improving the accuracy and reliability of determining that the vehicle does not meet the preset wake-up conditions.

[0067] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 2 The described parking safety monitoring method based on AI and IoT technologies can be applied to parking safety monitoring devices based on AI and IoT technologies. These devices may include monitoring equipment or a monitoring server, which may include a cloud server or a local server; this embodiment of the invention does not impose limitations. Figure 2 As shown, this parking safety monitoring method based on AI and IoT technologies may include the following operations: 201. Based on the multi-dimensional feature information of the detected object detected by the detection equipment corresponding to the vehicle, calculate the target judgment value corresponding to the vehicle.

[0068] 202. By determining the environmental information of the vehicle's environment, calculate the vehicle's adaptive judgment threshold regarding the environment.

[0069] In this embodiment of the invention, there is no order between step 202 and step 201. That is, step 202 can occur before step 201, after step 201, or simultaneously with step 201. This embodiment of the invention does not impose any limitations.

[0070] 203. Based on the target judgment value and the adaptive judgment threshold, determine whether the vehicle meets the preset wake-up conditions.

[0071] In this embodiment of the invention, when the judgment result of step 203 is yes, that is, when it is determined that the vehicle meets the wake-up condition, step 204 is triggered; when the judgment result of step 203 is no, the operation of steps 201-202 can be re-executed, or the process can be terminated. This embodiment of the invention does not limit the scope of the process.

[0072] 204. Control the vehicle to enter the wake-up state, and after the vehicle is woken up, control the safety monitoring equipment that matches the detection equipment to monitor the vehicle-related image data according to the location of the detection equipment.

[0073] In this embodiment of the invention, optionally, the safety monitoring device (such as a camera) matching the detection device can be the safety monitoring device closest to the detection device among all the safety monitoring devices corresponding to the vehicle. Compared to the traditional method of activating all cameras to take pictures, which results in excessive standby power consumption because only the camera in the direction in which an object approaches the vehicle can capture the image, this solution can activate the camera closest to the detection device (such as millimeter-wave radar) to take pictures, which not only reduces the vehicle's standby power consumption but also improves the accuracy and efficiency of image data monitoring. Further optionally, the safety monitoring device matching the detection device can also be one or more safety monitoring devices determined based on the vehicle model. Specifically, for low-computing-power vehicles, a lightweight splicing stream solution can be used to achieve basic monitoring with only a single camera; for high-computing-power vehicles, a multi-camera stream solution can be used to achieve comprehensive monitoring by multiple cameras.

[0074] In this embodiment of the invention, the image data may optionally include video data and / or image data, and the amount of image data can be obtained by setting the monitoring duration of the security monitoring device. For example, the camera can shoot a video of a preset duration (such as 20 seconds) when shooting. This embodiment of the invention does not limit this.

[0075] In this embodiment of the invention, for other descriptions of steps 201-204, please refer to the detailed description of steps 101-104 in Embodiment 1. These descriptions will not be repeated in this embodiment of the invention.

[0076] 205. Analyze the image data to determine whether the vehicle meets the preset warning conditions.

[0077] In this embodiment of the invention, optionally, the warning condition can be used to indicate that the proportion of video frames containing a preset type of foreground object in the video data is greater than a preset ratio value, or it can be used to indicate that the proportion of image data containing a preset type of foreground object in the image data is greater than a preset ratio value. The preset type of foreground object is the human body, but this embodiment of the invention does not limit the definition.

[0078] In this embodiment of the invention, when the judgment result of step 205 is yes, that is, when it is determined that the vehicle meets the warning conditions, step 206 is triggered; when the judgment result of step 205 is no, that is, when it is determined that the vehicle does not meet the warning conditions, the operation of steps 204-205 can be re-executed, or the process can be terminated. This embodiment of the invention does not impose any limitations.

[0079] 206. Send warning notifications to the vehicle owners whose vehicles are linked to the vehicle.

[0080] In this embodiment of the invention, optionally, the notification channel for sending warning notification information to the car owner can be an APP push channel or an SMS notification channel, and the APP push channel has a higher priority than the SMS notification channel. The notification process of the APP push channel can be to send an encrypted push through the car manufacturer's official APP, including the warning level, time, location and video summary (first frame + key frame), etc. The SMS notification channel can be to send an SMS with a short link (link pointing to the cloud video) when the APP does not respond.

[0081] Optionally, the communication protocol for sending warning notifications to vehicle owners can be either a CP / IP protocol supported by the T5G NSA / SA network, configured in the vehicle communication module, or an HTTPS protocol. The HTTPS protocol is used to compress the video stream using the H.265 video compression standard before uploading it to the cloud. The video format can be encapsulated in JSON format, for example: JSON encapsulates metadata: {"alert_id":"xxx","location":"GPS coordinates","risk_level":"high"}.

[0082] It is evident that implementation Figure 2The described parking safety monitoring method based on AI and IoT technologies can accurately calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object, and automatically determine the adaptive judgment threshold of the vehicle with respect to the environment based on the environmental information of the vehicle's location. Then, based on the target judgment value and the adaptive judgment threshold, it can accurately determine whether the vehicle meets the preset wake-up conditions. This allows the judgment threshold used as the basis for vehicle wake-up judgment to adaptively change with the changes in the vehicle's environment, making it more applicable to different scenarios and reducing the probability of false judgment, thereby reducing the probability of the vehicle being falsely woken up. Furthermore, after the vehicle is woken up, the matching safety monitoring equipment controlled by the detection equipment can perform corresponding safety monitoring operations on the vehicle, which can improve the matching degree between the detection equipment and the safety monitoring equipment with control requirements, thereby improving the accuracy and efficiency of controlling the safety monitoring equipment and maintaining a good safety monitoring effect. In addition, after the vehicle is awakened, the system can control the matching safety monitoring equipment to monitor the vehicle's image data based on the location of the detection equipment. This can improve the accuracy and efficiency of image data monitoring, thereby analyzing the image data to accurately determine whether the vehicle meets the preset warning conditions. If the vehicle meets the warning conditions, a warning notification will be sent to the vehicle owner who is associated with the vehicle in a timely and accurate manner.

[0083] In an optional embodiment, the analysis of image data in step 205 above to determine whether the vehicle meets the preset warning conditions may include: When the image data is video data, the video data is split to obtain multiple initial video frames corresponding to the video data; Each initial video frame is input into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame; Based on the object type of the foreground object indicated by all initial video frames, select all target video frames whose object type is a preset type from all initial video frames; Calculate the ratio between the number of target video frames and the number of initial video frames, and determine whether the ratio is greater than a preset ratio; When the ratio is determined to be greater than the preset ratio, the vehicle is determined to meet the preset warning conditions; When the ratio is determined to be less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions.

[0084] In this embodiment of the invention, specifically, a low-power NPU (Neural Processing Unit, i.e., embedded neural network processor) can be set up in the vehicle to perform video analysis. The NPU splits the video into multiple frames and inputs each video frame into a pre-trained classification model for analysis, thereby determining the type of foreground object in the video frame.

[0085] In this embodiment of the invention, optionally, the number of target video frames can be zero or an integer greater than zero. When the number of target video frames is zero, it indicates that the object type of the foreground object indicated by all video frames is not a preset type (such as a human body). When the number of target video frames is an integer greater than zero, it indicates that there is a foreground object of a preset type among the foreground objects indicated by all video frames. This embodiment of the invention does not impose any limitations.

[0086] For example, the classification model could be the YOLOv13 model, which significantly improves the detection capabilities of occlusion and small targets (such as pedestrians in the distance) by introducing hypergraph theory and modeling high-order semantic relationships between multiple targets. Furthermore, its Nano version achieves an mAP of 41.6% on the COCO dataset with a computational cost of only 6.4 GFLOPs, making it suitable for in-vehicle embedded deployment. In this way, the HyperGraph structure can handle occluded targets (such as pedestrians behind garage pillars), and its Nano version is compatible with in-vehicle NPUs. The training materials for the classification model can be generated by AI based on a limited amount of existing materials (such as video frames with manually labeled object types). This expands the amount of materials available. For example, Conditional Generative Adversarial Networks (CGANs) can be used to add conditional information such as category labels to the model input to generate images of a specified category. A CGAN could be MCMC-CGAN, which combines Markov chain Monte Carlo algorithms to filter high-quality samples in the latent space to improve the clarity and diversity of the generated images. This is applicable to labeled datasets such as MNIST and CelebA. Alternatively, a CGAN (GAN with an auxiliary classifier) ​​can be used to add a classification task to the discriminator, ensuring a strong correlation between the generated images and the labels. This is commonly used for biological image generation. The training data generation process may specifically include: obtaining basic data through public datasets, such as COCO-Person (e.g., 50,000 annotated human images); using ACGAN (Auxiliary Classifier GAN) to augment the basic data; and using MCMC-CGAN to sample the generated images and remove blurry / structurally abnormal samples (retaining the top 30% of clear samples). Furthermore, the classification model can be optimized for deployment. The optimization process can include: class pruning, reducing the original model from supporting 80 classes to only retaining the "person" class output layer, which helps reduce the computational load of the model; setting a frame analysis strategy, in which the frame sampling frequency is set (e.g., 100 frames are extracted from a 20-second video at 5fps); and setting an early warning condition in the frame analysis strategy, where the early warning condition is triggered when a human body is detected in 50 or more frames (i.e., the proportion is greater than or equal to 0.5).

[0087] As can be seen, this optional embodiment can split the monitored video data to obtain multiple initial video frames, and input each initial video frame into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame. This improves the accuracy and efficiency of classifying foreground objects in each frame of the video. Based on the object types of the foreground objects indicated by all initial video frames, it selects all target video frames with a preset object type from all initial video frames, improving the accuracy and efficiency of selecting target video frames belonging to the preset type. Subsequently, it calculates the ratio between the number of target video frames and the number of initial video frames, and determines whether the ratio is greater than a preset ratio. When the ratio is greater than the preset ratio, it is determined that the vehicle meets the preset warning conditions, improving the accuracy and reliability of determining whether the vehicle meets the preset warning conditions. When the ratio is less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions, improving the accuracy and reliability of determining whether the vehicle does not meet the preset warning conditions.

[0088] For example, such as Figure 3 As shown, Figure 3 This is a flowchart illustrating another parking safety monitoring method based on AI and IoT technologies disclosed in an embodiment of the present invention. The parking safety monitoring method based on AI and IoT technologies may include the following steps: S1 determines whether the vehicle meets the wake-up conditions by acquiring detection information from the millimeter-wave radar on the vehicle.

[0089] After being activated, the vehicle's camera system uses the millimeter-wave radar to control the corresponding camera to start and take pictures.

[0090] S3 analyzes the captured video to determine whether it meets the warning conditions.

[0091] S4 sends a warning notification to the vehicle owner when the warning conditions are met.

[0092] Specifically, step S1 may include the following steps: S10 calculates the target's velocity, range, micro-Doppler characteristics, and RCS value based on the detection information; S20, the judgment value is calculated based on speed, distance, micro-Doppler characteristics and RCS value; S30: Compare the judgment value with the adaptive judgment threshold to determine whether the wake-up condition is met.

[0093] The process of determining the adaptive judgment threshold in step S30 includes: The first step is to determine the vehicle's environmental type based on its location; The second step is to obtain the vehicle's wake-up record; The third step is to calculate an adaptive judgment threshold based on the environment type and wake-up record.

[0094] The method described above in this invention not only utilizes the ranging results of millimeter-wave radar, but also combines speed, micro-Doppler features, and RCS value to calculate the judgment value. Compared with relying solely on distance for judgment, it can effectively reduce the probability of false wake-up. Furthermore, the judgment threshold is not a fixed threshold, but can adaptively change based on the vehicle's environmental type and wake-up records. This makes it more applicable to different scenarios and can effectively reduce the probability of false wake-up.

[0095] For example, in order to implement the above method, the present invention also provides a parking safety monitoring device, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of the structure of a parking safety monitoring device disclosed in an embodiment of the present invention. The parking safety monitoring device may include a millimeter-wave radar, an onboard SOC, a camera, and a communication module; wherein: Millimeter-wave radar is used to acquire detection information in the direction of installation and transmit the detection information to the vehicle's SOC; The vehicle-mounted SOC is used to determine whether the vehicle meets the wake-up conditions by acquiring the detection information of the millimeter-wave radar on the vehicle, and to send control shooting commands to the camera according to the position of the millimeter-wave radar corresponding to the detection information when the vehicle meets the wake-up conditions. The camera is used to capture video in the direction of installation after receiving a shooting command, and to transmit the acquired video to the vehicle SOC. The vehicle-mounted SOC is also used to analyze the captured video to determine whether the warning conditions are met. When the warning conditions are met, the control communication module sends a warning notification to the vehicle owner.

[0096] The communication protocol used between the various devices in the parking safety monitoring system is as follows: The communication protocol between the millimeter-wave radar and the vehicle-mounted SOC can be CAN / CAN FD protocol, which has the characteristics of low latency (millisecond level), anti-interference, and multi-node networking. The communication protocol between the camera and the vehicle-mounted SOC can be the MIPI CSI-2 + D-PHY / C-PHY protocol, which features high bandwidth (1–10Gbps), dynamic rate adaptation, and I²C control channel. The communication protocol between the vehicle SOC and the communication module can be TCP / IP, which can be used to run vehicle Ethernet. The communication protocol for coordinating the internal subsystems of an onboard SOC can be the SOME / IP + AUTOSAR protocol.

[0097] Example 3 Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a parking safety monitoring device based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 5 The described parking safety monitoring device based on AI and IoT technologies may include monitoring equipment or a monitoring server, wherein the monitoring server may include a cloud server or a local server, and the embodiments of the present invention are not limited thereto. Figure 5 As shown, the parking safety monitoring device based on AI and IoT technologies may include: The calculation module 301 is used to calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle.

[0098] The calculation module 301 is also used to calculate the vehicle's adaptive judgment threshold about the environment by determining the environmental information of the environment in which the vehicle is located.

[0099] The judgment module 302 is used to determine whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold.

[0100] The control module 303 is used to control the vehicle to enter the wake-up state when the judgment module 302 determines that the vehicle meets the wake-up conditions, and after the vehicle is woken up, control the safety monitoring equipment matched with the detection equipment to perform corresponding safety monitoring operations on the vehicle.

[0101] It is evident that implementation Figure 5The described parking safety monitoring device, based on AI and IoT technologies, can accurately calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object, and automatically determine the adaptive judgment threshold of the vehicle with respect to the environment based on the environmental information of the vehicle's location. Then, based on the target judgment value and the adaptive judgment threshold, it accurately determines whether the vehicle meets the preset wake-up conditions. This allows the judgment threshold used as the basis for vehicle wake-up judgment to adaptively change with the changes in the vehicle's environment, making it more suitable for different scenarios and reducing the probability of false judgments, thereby reducing the probability of the vehicle being falsely woken up. Furthermore, after the vehicle is woken up, the device controls the matching safety monitoring equipment to perform corresponding safety monitoring operations on the vehicle, which can improve the matching degree between the detection equipment and the safety monitoring equipment with control requirements, thereby improving the accuracy and efficiency of controlling the safety monitoring equipment and maintaining a good safety monitoring effect.

[0102] In an optional embodiment, the calculation module 301 calculates the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle, including the following specific methods: Obtain detection information from the detection equipment corresponding to the vehicle, including at least one of the starting frequency bandwidth and the frequency sweep period; Based on the detection information, calculate the multi-dimensional feature information of the detected object detected by the detection device. The detected object is the object within the preset range where the vehicle is located. The multi-dimensional feature information includes multiple sub-class feature information, and all sub-class feature information includes at least two of the following: speed, distance, micro-Doppler features, and RCS value. For each subclass feature information, calculate the judgment value of the subclass feature information and determine the weight coefficient corresponding to the subclass feature information; The target judgment value corresponding to the vehicle is calculated based on the judgment values ​​of all subclass feature information and the weight coefficients corresponding to each subclass feature information.

[0103] As can be seen, this optional embodiment can calculate the multi-dimensional feature information (including multiple sub-class feature information) of the detected object detected by the detection device based on the detection information of the vehicle-related detection device, thereby calculating the judgment value of each sub-class feature information and determining the weight coefficient corresponding to that sub-class feature information. Based on the judgment values ​​of all sub-class feature information and the weight coefficient corresponding to each sub-class feature information, the target judgment value corresponding to the vehicle can be accurately calculated, which can improve the accuracy and reliability of the calculation of the target judgment value corresponding to the vehicle. This is beneficial to improving the accuracy of the judgment of whether the vehicle meets the wake-up conditions based on the target judgment value, and thus helps to reduce the probability of the vehicle being falsely woken up.

[0104] In this optional embodiment, as an optional implementation method, the specific way in which the calculation module 301 calculates the judgment value of the subclass feature information includes: When the subclass feature information is a micro-Doppler feature, the phase timing signal of the detection device is determined based on the detection information; Perform a short-time Fourier transform on the phase-time signal to obtain the time spectrum; and extract the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time spectrum as the main modulation frequency; If the main modulation frequency is greater than or equal to the first preset frequency threshold and less than or equal to the second preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the first judgment value, and the first preset frequency threshold is less than the second preset frequency threshold. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the second judgment value, and the second preset frequency threshold is less than the third preset frequency threshold. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the third judgment value, the first judgment value is greater than the second judgment value, and the second judgment value is greater than the third judgment value.

[0105] As can be seen, this optional implementation can automatically determine the phase timing signal of the detection device based on the detection information for micro-Doppler features, and perform a short-time Fourier transform on the phase timing signal to obtain a time-frequency spectrum. It then extracts the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time-frequency spectrum as the main modulation frequency. This improves the accuracy of extracting the main modulation frequency used as the basis for determining the judgment value of the micro-Doppler feature. If the main modulation frequency is greater than or equal to a first preset frequency threshold and less than or equal to a second preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the first judgment value. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to a third preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the second judgment value. If the main modulation frequency is less than the first preset frequency threshold or greater than the third preset frequency threshold, the judgment value of the micro-Doppler feature is determined as the third judgment value. By comparing the results of multiple preset frequency thresholds with the main modulation frequency, the accuracy and reliability of determining the judgment value of the micro-Doppler feature can be improved, thereby enhancing the accuracy and reliability of subsequently calculating the target judgment value corresponding to the vehicle based on the judgment value of the micro-Doppler feature.

[0106] In another optional embodiment, the calculation module 301 calculates the vehicle's adaptive judgment threshold regarding the environment by determining the environmental information of the vehicle's environment, specifically in the following ways: Obtain the vehicle's location information and wake-up record information; Based on the vehicle's location information, determine the environmental type of the vehicle's location, and the corresponding risk level of the environmental type. Based on the risk level corresponding to the environmental type, a benchmark judgment value is determined for the corresponding environmental type. The risk level is negatively correlated with the benchmark judgment value. Based on the vehicle's wake-up record information, the wake-up record feature value of the vehicle is calculated. The wake-up record feature value is calculated under the condition that the vehicle is in the same location information. Based on the baseline judgment value corresponding to the environment type and the vehicle's wake-up record feature value, calculate the vehicle's adaptive judgment threshold regarding the environment.

[0107] As can be seen, this optional embodiment can automatically determine the environment type of the vehicle's environment based on the acquired vehicle positioning information, and determine the benchmark judgment value corresponding to the environment type based on the risk level corresponding to the environment type. It can improve the accuracy and efficiency of determining the benchmark judgment value based on the risk level corresponding to the environment type, and can quickly and accurately calculate the vehicle's wake-up record feature value based on the vehicle's wake-up record information. Thus, based on the benchmark judgment value and the wake-up record feature value, it can calculate the vehicle's adaptive judgment threshold with respect to the environment, thereby improving the accuracy and reliability of calculating the adaptive judgment threshold.

[0108] In another optional embodiment, the specific method by which the judging module 302 judges whether the vehicle meets the preset wake-up conditions based on the target judging value and the adaptive judging threshold includes: The comparison results are obtained by comparing the target judgment value with the adaptive judgment threshold; When the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it is determined that the vehicle meets the preset wake-up conditions; When the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it is determined that the vehicle does not meet the preset wake-up conditions.

[0109] As can be seen, this optional embodiment can compare the target judgment value and the adaptive judgment threshold, and when the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it determines that the vehicle meets the preset wake-up conditions, thereby improving the accuracy and reliability of determining that the vehicle meets the preset wake-up conditions. Conversely, when the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it determines that the vehicle does not meet the preset wake-up conditions, thereby improving the accuracy and reliability of determining that the vehicle does not meet the preset wake-up conditions.

[0110] In another optional embodiment, the control module 303 controls a safety monitoring device matching the detection device to perform corresponding safety monitoring operations on the vehicle, including: Based on the location of the detection equipment, control the matching safety monitoring equipment to monitor vehicle-related image data. The image data is analyzed to determine whether the vehicle meets the preset warning conditions; When a vehicle is determined to meet the warning conditions, a warning notification is sent to the owner of the vehicle.

[0111] As can be seen, this optional embodiment can, after the vehicle is awakened, control the safety monitoring equipment matched with the detection equipment to monitor the vehicle-related image data according to the location of the detection equipment. This can improve the accuracy and efficiency of image data monitoring, thereby analyzing the image data to accurately determine whether the vehicle meets the preset warning conditions. If the vehicle meets the warning conditions, a warning notification message is accurately and promptly sent to the vehicle owner who is associated with the vehicle.

[0112] In this optional embodiment, as an optional implementation, the image data includes video data and / or image data. Furthermore, the control module 303 analyzes the image data to determine whether the vehicle meets preset warning conditions, including: When the image data is video data, the video data is split to obtain multiple initial video frames corresponding to the video data; Each initial video frame is input into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame; Based on the object type of the foreground object indicated by all initial video frames, select all target video frames whose object type is a preset type from all initial video frames; Calculate the ratio between the number of target video frames and the number of initial video frames, and determine whether the ratio is greater than a preset ratio; When the ratio is determined to be greater than the preset ratio, the vehicle is determined to meet the preset warning conditions; When the ratio is determined to be less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions.

[0113] As can be seen, this optional implementation can split the monitored video data to obtain multiple initial video frames, and input each initial video frame into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame. This improves the accuracy and efficiency of classifying foreground objects in each frame of the video. Based on the object types of the foreground objects indicated by all initial video frames, all target video frames with the object type of the preset type are selected from all initial video frames, improving the accuracy and efficiency of selecting target video frames belonging to the preset type. Then, the ratio between the number of target video frames and the number of initial video frames is calculated, and it is determined whether the ratio is greater than a preset ratio. When the ratio is determined to be greater than the preset ratio, it is determined that the vehicle meets the preset warning conditions, improving the accuracy and reliability of determining whether the vehicle meets the preset warning conditions. When the ratio is determined to be less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions, improving the accuracy and reliability of determining whether the vehicle does not meet the preset warning conditions.

[0114] Example 4 Please see Figure 6 , Figure 6 This is a schematic diagram of another parking safety monitoring device based on AI and IoT technologies disclosed in an embodiment of the present invention. Figure 6 As shown, the parking safety monitoring device based on AI and IoT technologies may include: Memory 401 storing executable program code; Processor 402 coupled to memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the parking safety monitoring method based on AI and IoT technologies described in Embodiment 1 or Embodiment 2 of the present invention.

[0115] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the parking safety monitoring method based on AI and IoT technologies described in Embodiment 1 or Embodiment 2 of this invention.

[0116] Example 6 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the parking safety monitoring method based on AI and IoT technologies described in Embodiment 1 or Embodiment 2.

[0117] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0118] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0119] Finally, it should be noted that the parking safety monitoring method and device based on AI and IoT technologies disclosed in the embodiments of this invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this invention.

Claims

1. A parking safety monitoring method based on AI and IoT technologies, characterized in that, The method includes: Based on the multi-dimensional feature information of the detected object detected by the detection equipment corresponding to the vehicle, the target judgment value corresponding to the vehicle is calculated. By determining the environmental information of the environment in which the vehicle is located, an adaptive judgment threshold for the vehicle with respect to the environment is calculated; Based on the target judgment value and the adaptive judgment threshold, determine whether the vehicle meets the preset wake-up conditions; When it is determined that the vehicle meets the wake-up condition, the vehicle is controlled to enter the wake-up state, and after the vehicle is woken up, the safety monitoring device matching the detection device is controlled to perform corresponding safety monitoring operations on the vehicle. The step of determining the environmental information of the vehicle's environment and calculating the vehicle's adaptive judgment threshold regarding the environment includes: Obtain the vehicle's location information and the vehicle's wake-up record information; Based on the vehicle's location information, the environmental type of the environment in which the vehicle is located is determined, and the environmental type has a corresponding level of risk. Based on the risk level corresponding to the environmental type, a baseline judgment value corresponding to the environmental type is determined, and the risk level is negatively correlated with the baseline judgment value; Based on the vehicle's wake-up record information, the wake-up record feature value of the vehicle is calculated, and the wake-up record feature value is calculated under the condition that the vehicle is under the same location information; Based on the baseline judgment value corresponding to the environment type and the wake-up record feature value of the vehicle, the adaptive judgment threshold of the vehicle with respect to the environment is calculated.

2. The parking safety monitoring method based on AI and IoT technologies according to claim 1, characterized in that, The step of calculating the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle includes: Obtain detection information from the detection device corresponding to the vehicle, wherein the detection information includes at least one of the starting frequency bandwidth and the frequency sweep period; Based on the detection information, the multi-dimensional feature information of the detected object is calculated. The detected object is an object within a preset range where the vehicle is located. The multi-dimensional feature information includes multiple sub-class feature information, and all the sub-class feature information includes at least two of the following: speed, distance, micro-Doppler features, and RCS value. For each of the subclass feature information, calculate the judgment value of the subclass feature information and determine the weight coefficient corresponding to the subclass feature information; The target judgment value corresponding to the vehicle is calculated based on the judgment values ​​of all the sub-class feature information and the weight coefficient corresponding to each sub-class feature information.

3. The parking safety monitoring method based on AI and IoT technologies according to claim 2, characterized in that, The calculation of the judgment value of the subclass feature information includes: When the subclass feature information is the micro-Doppler feature, the phase timing signal of the detection device is determined based on the detection information; Perform a short-time Fourier transform on the phase timing signal to obtain a time spectrum; and extract the frequency of the maximum energy peak from the multiple energy peak frequencies indicated by the time spectrum as the main modulation frequency; If the main modulation frequency is greater than or equal to the first preset frequency threshold and less than or equal to the second preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the first judgment value, where the first preset frequency threshold is less than the second preset frequency threshold. If the main modulation frequency is greater than the second preset frequency threshold and less than or equal to the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the second judgment value, and the second preset frequency threshold is less than the third preset frequency threshold. If the main modulation frequency is less than the first preset frequency threshold, or greater than the third preset frequency threshold, then the judgment value of the micro-Doppler feature is determined to be the third judgment value, where the first judgment value is greater than the second judgment value, and the second judgment value is greater than the third judgment value.

4. The parking safety monitoring method based on AI and IoT technologies according to any one of claims 1-3, characterized in that, The step of determining whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold includes: The comparison result is obtained by comparing the target judgment value with the adaptive judgment threshold; When the comparison result indicates that the target judgment value is greater than the adaptive judgment threshold, it is determined that the vehicle meets the preset wake-up conditions; When the comparison result indicates that the target judgment value is less than or equal to the adaptive judgment threshold, it is determined that the vehicle does not meet the preset wake-up conditions.

5. The parking safety monitoring method based on AI and IoT technology according to any one of claims 1-3, characterized in that, The safety monitoring equipment, which controls the detection equipment to perform corresponding safety monitoring operations on the vehicle, includes: Based on the location of the detection device, control a matching safety monitoring device to monitor the vehicle-related image data. The image data is analyzed to determine whether the vehicle meets the preset warning conditions; When it is determined that the vehicle meets the warning conditions, a warning notification is sent to the owner of the vehicle.

6. The parking safety monitoring method based on AI and IoT technologies according to claim 5, characterized in that, The image data includes video data and / or image data; And, the analysis of the image data to determine whether the vehicle meets the preset warning conditions includes: When the image data is video data, the video data is split to obtain multiple initial video frames corresponding to the video data; Each initial video frame is input into a pre-trained classification model for classification to obtain the object type of the foreground object indicated by the initial video frame; Based on the object type of the foreground object indicated by all the initial video frames, select all target video frames whose object type is a preset type from all the initial video frames; Calculate the ratio between the number of target video frames and the number of initial video frames, and determine whether the ratio is greater than a preset ratio; When it is determined that the ratio is greater than the preset ratio, the vehicle is determined to meet the preset warning conditions; When it is determined that the ratio is less than or equal to the preset ratio, it is determined that the vehicle does not meet the preset warning conditions.

7. A parking safety monitoring device based on AI and IoT technologies, characterized in that, The device is used to perform the parking safety monitoring method based on AI and IoT technologies as described in any one of claims 1-6, the device comprising: The calculation module is used to calculate the target judgment value corresponding to the vehicle based on the multi-dimensional feature information of the detected object detected by the detection device corresponding to the vehicle. The calculation module is also used to calculate the adaptive judgment threshold of the vehicle with respect to the environment by determining the environmental information of the environment in which the vehicle is located; The judgment module is used to determine whether the vehicle meets the preset wake-up conditions based on the target judgment value and the adaptive judgment threshold. The control module is used to control the vehicle to enter the wake-up state when the judgment module determines that the vehicle meets the wake-up conditions, and after the vehicle is woken up, control the safety monitoring device matched with the detection device to perform corresponding safety monitoring operations on the vehicle.

8. A parking safety monitoring device based on AI and IoT technologies, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the parking safety monitoring method based on AI and IoT technologies as described in any one of claims 1-6.

9. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the parking safety monitoring method based on AI and IoT technologies as described in any one of claims 1-6.

Images