Scene recognition method and device, equipment and storage medium

Abstract

The invention discloses a scene recognition method and apparatus, a device and a storage medium. The method comprises the steps of obtaining first real-time information; wherein the first real-time information at least comprises one or more of the following items: sensor information; cell number change information; cell signal intensity change information; searching information by a satellite; a cell monitoring result; and performing scene recognition of the basement scene based on the first real-time information to obtain a scene recognition result.

Description

Technical Field

[0001] This invention relates to the field of scene recognition technology, and in particular to a scene recognition method, apparatus, device, and storage medium. Background Technology

[0002] Scene recognition technology is an important means for intelligent devices to automatically determine their location in different environments, and it is widely used in navigation, positioning, and personalized services. However, current common scene recognition technologies suffer from poor accuracy in practical applications. Summary of the Invention

[0003] This application provides a scene recognition method, apparatus, device, and storage medium that can achieve more accurate scene recognition.

[0004] The technical solution of this application embodiment is implemented as follows:

[0005] In a first aspect, embodiments of this application provide a scene recognition method, the method comprising:

[0006] Acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results;

[0007] Scene recognition is performed on the underground parking lot scene based on the first real-time information to obtain the scene recognition results.

[0008] Secondly, embodiments of this application provide a scene recognition device, which includes:

[0009] The acquisition unit is used to acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results;

[0010] The recognition unit is used to perform scene recognition of the underground parking lot scene based on the first real-time information and obtain the scene recognition result.

[0011] Thirdly, embodiments of this application provide an electronic device, which includes a processor and a memory storing processor-executable instructions, wherein when the instructions are executed by the processor, the method as described in the first aspect is implemented.

[0012] Fourthly, embodiments of this application provide a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the method described in the first aspect above.

[0013] This application provides a scene recognition method, apparatus, device, and storage medium to acquire first real-time information. The first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; and scene recognition of a parking garage scene based on the first real-time information to obtain a scene recognition result. Therefore, in the embodiments of this application, one or more types of real-time information related to the scene recognition device itself can be collected, such as sensor information, cell number change information, cell signal strength change information, satellite search information, etc. By fusing multiple types of real-time information, a comprehensive judgment and full recognition of the parking garage scene can be performed, thereby achieving multimodal, high-precision scene recognition and improving the accuracy and universality of scene recognition. Attached Figure Description

[0014] Figure 1 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0015] Figure 2 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0016] Figure 3 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0017] Figure 4 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0018] Figure 5 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0019] Figure 6 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0020] Figure 7 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0021] Figure 8 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0022] Figure 9 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0023] Figure 10 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application.

[0024] Figure 11 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0025] Figure 12 This is a schematic diagram illustrating the implementation of the scene recognition method proposed in this application.

[0026] Figure 13 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application.

[0027] Figure 14 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0028] Figure 15 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application.

[0029] Figure 16 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0030] Figure 17 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in this application embodiment;

[0031] Figure 18 This is a schematic diagram of the composition structure of the scene recognition device proposed in the embodiments of this application;

[0032] Figure 19 This is a schematic diagram of the composition structure of the electronic device proposed in the embodiments of this application. Detailed Implementation

[0033] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are only for explaining the differences between the application and the original application, and are not intended to limit the application. Furthermore, it should be noted that, for ease of description, only the parts that differ from the relevant application are shown in the accompanying drawings.

[0034] Scene recognition technology is a crucial tool for intelligent devices to automatically determine their location in different environments, and it is widely used in navigation, positioning, and personalized services. With the development of mobile communication and sensing technologies, scene recognition methods are constantly evolving to improve accuracy and applicability.

[0035] In related technologies, during the scene recognition process, one approach is for the mobile phone to determine the scene recognition result based on data provided by a third-party service. Another approach is for the mobile phone to analyze changes in gravity direction using its built-in gravity sensor to determine whether it is in an underground space, thereby performing underground parking scene recognition. External data support can also be used to improve the recognition effect.

[0036] However, on the one hand, the data coverage of third-party services is limited, which makes it impossible to meet the recognition coverage requirements. At the same time, the strong dependence on third-party services reduces the independence of the system. On the other hand, gravity sensors are prone to misjudgment at certain angles, such as when the phone is close to vertical, the error will be very large, which will lead to low overall recognition accuracy of scene recognition.

[0037] It is evident that current common scene recognition technologies suffer from problems such as small recognition coverage, insufficient accuracy, and poor system independence in practical applications, making it difficult to meet the high requirements for underground parking scene recognition in complex environments.

[0038] To address the aforementioned issues, embodiments of this application provide a scene recognition method, apparatus, device, and storage medium for acquiring first real-time information. This first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; and scene recognition of a parking garage scene based on the first real-time information to obtain a scene recognition result. Therefore, embodiments of this application can collect one or more types of real-time information related to the scene recognition device itself, such as sensor information, cell number change information, cell signal strength change information, and satellite search information. By fusing multiple types of real-time information, a comprehensive judgment and full recognition of the parking garage scene can be achieved, thereby realizing multimodal, high-precision scene recognition and improving the accuracy and universality of scene recognition.

[0039] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0040] One embodiment of this application provides a scene recognition method, which can be applied to a scene recognition device or electronic device, and can also be applied to any terminal that includes a scene recognition device or electronic device.

[0041] It is understood that the scene recognition method proposed in this application embodiment may include a real-time recognition scheme for underground parking scenes. By integrating multiple types of real-time information, it can make comprehensive judgments and full recognition of underground parking scenes, thereby achieving multimodal and high-precision scene recognition and improving the accuracy and universality of scene recognition.

[0042] The following description uses a scene recognition device as an example to illustrate the scene recognition method proposed in this application.

[0043] Furthermore, in the embodiments of this application, Figure 1 This is a schematic diagram illustrating the implementation process of the scene recognition method proposed in the embodiments of this application, such as... Figure 1As shown, the scene recognition method may include the following steps:

[0044] Step 101: Obtain first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results.

[0045] In the embodiments of this application, the scene recognition device may first collect one or more types of information in real time to obtain first real-time information.

[0046] In the embodiments of this application, the scene recognition device can be a mobile terminal or an in-vehicle terminal. When implementing the scene recognition method proposed in this application, the scene recognition device can be located in the vehicle equipment, or it can be loaded into the vehicle equipment; this application does not impose specific limitations.

[0047] In the embodiments of this application, the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results.

[0048] In other words, in the embodiments of this application, the scene recognition device can collect different types of information and data in real time, and there is no specific limitation on the specific type and content of the first real-time information acquired.

[0049] In some embodiments, sensor information can be used to determine the motion state of the scene recognition device, such as whether the scene recognition device is stationary, moving uphill or downhill, or tilted. The sensor information includes, but is not limited to, acceleration data and gyroscope data.

[0050] In some embodiments, when acquiring the first real-time information, acceleration data can be acquired through an accelerometer; or gyroscope data can be acquired through a gyroscope.

[0051] An accelerometer is a hardware module used to detect linear acceleration changes of a scene recognition device in three-dimensional space. The accelerometer measures the acceleration values ​​of the device along the X, Y, and Z axes, typically expressed in m / s². In this application, the accelerometer is used to capture dynamic changes during the movement of the scene recognition device, such as acceleration changes during uphill or downhill movement, thereby assisting in determining whether the device has entered or left a parking lot scene.

[0052] A gyroscope is a sensor used to detect changes in the rotation angle of a device around its own axis, typically outputting angular velocity data in radians per second (rad / s). Gyroscopes are used to detect changes in the device's attitude, such as pitch, yaw, and roll angles, thereby helping the system determine whether the scene recognition device is stationary or moving. In this application, gyroscope data provides information about the rotational state of the scene recognition device, helping to distinguish whether the movement is walking, driving, or stationary.

[0053] In some embodiments, when identifying a parking garage scene, a gyroscope and an accelerometer can be used together to form an inertial measurement unit (IMU) for more accurate determination of whether the environment is on an incline or slope. Accelerometer data reflects the acceleration state of the device at different points in time and can be used to calculate information such as the motion trajectory, velocity changes, and attitude changes of the scene recognition device. Gyroscope data reflects the rotational state of the device. When identifying a parking garage scene, combining acceleration and gyroscope data, and using a fusion algorithm (such as Kalman filtering) to estimate the device's attitude and motion trajectory, can improve the accuracy and robustness of the identification.

[0054] Therefore, in the embodiments of this application, a gyroscope and an accelerometer can be combined to acquire real-time data, providing more accurate real-time data for subsequent scene recognition processing. This avoids the low accuracy caused by sudden movement interference with the accelerometer (velocity sensor), and also solves the problem of the gyroscope's error increasing over time.

[0055] For example, in some embodiments, the scene recognition device may be configured with an inertial measurement unit (IMU). An IMU is a sensor module integrating an accelerometer and a gyroscope, used to collect motion state information of the scene recognition device in real time, i.e., to collect sensor information.

[0056] For example, in some embodiments, in addition to accelerometers and gyroscopes, the IMU configured in the scene recognition device may also include other detection modules of any type and function. For example, the IMU may include at least one or more of the following: accelerometer, gyroscope, magnetometer, distance sensor, gravity sensor, etc.

[0057] For example, in some embodiments, when a vehicle enters an underground parking garage, it will experience a significant downhill process. At this time, the IMU configured in the scene recognition device in the vehicle can accurately capture this change and provide a basis for subsequent judgment.

[0058] In some embodiments, cell number change information can be used to determine the changes in the cellular network signal corresponding to the scene recognition device. Specifically, the cell number change information can be used to deduce whether someone is entering or leaving the underground parking garage.

[0059] In some embodiments, cell signal strength change information can be used to determine the changes in the cellular network signal corresponding to the scene recognition device. Specifically, the cell signal strength change information can be used to deduce whether someone is entering or leaving a parking garage.

[0060] In some embodiments, a cellular cell is the basic coverage unit in a mobile communication network, and each cell is managed by a base station.

[0061] For example, in some embodiments, the scene recognition device may have cellular communication capabilities.

[0062] For example, in some embodiments, when a vehicle enters an underground parking garage, the number of serving cells may decrease and the signal strength of the serving cells may also decrease significantly due to the shielding effect of the garage structure on the signal. At this time, the underground parking garage scene can be further judged and deduced by the number of cells and / or the fluctuation of cell signals (cell number change information and / or cell signal strength change information) corresponding to the scene recognition device in the vehicle, as well as other types of cell information that are monitored.

[0063] In some embodiments, satellite search information can be used to determine whether the scene recognition device can receive satellite signals.

[0064] For example, in some embodiments, the scene recognition device may be configured with a Global Positioning System (GPS) module. By determining the satellite signals received by the configured GPS module (i.e., satellite search signals), it is possible to determine and deduce whether the scene recognition device is located in a basement.

[0065] For example, in some embodiments, the satellite search information of the scene recognition device can be monitored in real time, such as whether multiple satellite signals can be received stably, or whether satellite signals cannot be received for a long time, or whether the satellite signal quality is poor, so as to further judge and deduce the underground parking scene based on the determined satellite search information.

[0066] In some embodiments, the cell monitoring results may include information about the current serving cell and its neighboring cells monitored by the scene recognition device, including cell ID, signal strength, number of handovers, etc.

[0067] For example, in some embodiments, by determining the cell monitoring results corresponding to the scene recognition device and combining them with a pre-established preset fence cell, it is possible to determine and deduce whether the scene recognition device is in the underground parking lot.

[0068] Therefore, in the embodiments of this application, the scene recognition device may include smartphones, vehicle terminals, etc., equipped with IMU sensors, GPS modules, and cellular communication functions.

[0069] Step 102: Based on the first real-time information, perform scene recognition of the underground parking lot scene and obtain the scene recognition result.

[0070] In the embodiments of this application, after obtaining one or more of the first real-time information, including sensor information, cell number change information, cell signal strength change information, satellite search information, and cell monitoring results, scene recognition of the underground parking lot scene can be further performed based on the first real-time information to determine the corresponding scene recognition result.

[0071] In the embodiments of this application, the scene recognition results may include, but are not limited to: underground parking scene, non-underground parking scene, scene of entering underground parking, scene of leaving underground parking, etc.

[0072] In some embodiments, the scene recognition device can determine the underground parking scene in real time by fusing multiple sensing data such as IMU, GPS and cellular network, thereby enabling efficient and accurate recognition of the underground parking scene without relying on third-party services.

[0073] In the embodiments of this application, when performing scene recognition of the underground parking scene based on the first real-time information and obtaining the scene recognition result, real-time status information can be determined based on sensor information; then, based on the real-time status information, cell number change information and / or cell signal strength change information, the scene recognition result is determined.

[0074] In some embodiments, real-time status information is a status description dynamically generated based on sensor information, used to reflect the current physical environment or behavioral pattern of the scene recognition device. Real-time status information may include whether the device is stationary, whether it is going uphill or downhill, or whether it is traveling on a level surface.

[0075] For example, in some embodiments, the inertial measurement unit determines whether the device is going uphill or downhill by fusing data from the accelerometer and gyroscope. When the scene recognition device is going uphill, it usually means that it is exiting the ramp area of ​​an underground parking garage.

[0076] In some embodiments, by using real-time status information, the scene recognition device can more accurately determine whether a specific scene, such as a parking garage, has been entered. Specifically, determining real-time status information based on sensor data improves the accuracy of scene recognition, thereby more effectively determining whether an entry into a parking garage scene has occurred, and consequently enhancing the reliability of subsequent network optimization and experience prediction.

[0077] In the embodiments of this application, when determining real-time state information based on sensor information, the real-time dynamic and static state is determined based on acceleration data and / or gyroscope data; when the real-time dynamic and static state is non-static, the real-time pitch angle is determined based on acceleration data and / or gyroscope data; and the real-time state information is determined based on the real-time pitch angle.

[0078] In some embodiments, by fusing data from both acceleration and gyroscope sensors, it is possible to more accurately determine whether a device is in a stationary (static) or moving (non-static) state. For example, when going up or down a slope, both the accelerometer and gyroscope will detect significant dynamic changes, and the system will classify it as non-static. Furthermore, fusing acceleration and gyroscope data to determine real-time motion and static states improves the accuracy of motion state recognition for scene identification devices, avoids misjudgments due to errors from a single sensor, and thus enhances the reliability of subsequent state determinations.

[0079] In some embodiments, the pitch angle refers to the tilt angle of the scene recognition device relative to the horizontal plane, typically with the X-axis as a reference, representing the degree of upward or downward tilt of the front of the scene recognition device. When the scene recognition device is in a non-static state, the current pitch angle can be further calculated using data from the accelerometer and gyroscope. For example, when going uphill or downhill, the real-time pitch angle of the scene recognition device will change significantly, and by continuously monitoring the trend of the real-time pitch angle change, it can be determined whether it is going uphill or downhill.

[0080] In the embodiments of this application, by calculating the real-time pitch angle in a non-static state, the spatial attitude changes of the scene recognition device can be captured more accurately, thereby providing key basis for subsequent judgment of the scene (such as entering or leaving the underground parking lot), and thus enabling a more intelligent scene recognition function.

[0081] In some embodiments, real-time status information can be inferred by the scene recognition device based on the current posture and movement, such as whether it is entering or leaving a parking garage. Determining real-time status information through real-time pitch angle can effectively improve the perception of the environment, thereby enabling more accurate identification of the scene.

[0082] In the embodiments of this application, real-time dynamic and static states are determined by fusing acceleration data and gyroscope data, and the real-time pitch angle is further calculated in non-static cases. Finally, real-time state information is determined based on the real-time pitch angle. This improves the accuracy of recognizing the motion state and spatial attitude of the scene recognition device.

[0083] In the embodiments of this application, when determining the scene recognition result based on real-time status information, cell number change information and / or cell signal strength change information, if the real-time status information is an uphill state and the cell number change information is an increase in the number, the scene recognition result is determined to be a scene of leaving the underground parking lot.

[0084] In the embodiments of this application, when determining the scene recognition result based on real-time status information, cell number change information and / or cell signal strength change information, if the real-time status information is uphill and the cell signal strength change information is signal strength enhancement, the scene recognition result is determined to be a scene of leaving the underground parking lot.

[0085] In the embodiments of this application, when determining the scene recognition result based on real-time status information, cell number change information, and / or cell signal strength change information, if the real-time status information indicates a downhill state and the cell number change information indicates a decrease in the number of cells, the scene recognition result is determined to be an entry into the underground parking garage; and / or,

[0086] In the embodiments of this application, when determining the scene recognition result based on real-time status information, cell number change information and / or cell signal strength change information, if the real-time status information is a downhill state and the cell signal strength change information is a signal strength weakening, the scene recognition result is determined to be an entry into the underground parking lot.

[0087] In some embodiments, sensor information refers to data collected by the inertial measurement unit (IMU) built into the scene recognition device, including but not limited to information output by accelerometers and gyroscopes. This data is used to determine whether the scene recognition device is in motion and the attitude changes of the scene recognition device. For example, when a user drives into an underground parking garage, the scene recognition device on the vehicle will detect the acceleration changes of the vehicle moving up and down the ramp, and calculate the current attitude angle by combining the gyroscope data, thereby determining whether the device is in a slope state.

[0088] In some embodiments, cell count change information refers to the change in the number of cellular network cells connected to the scene recognition device within a certain time window. For example, when entering an underground parking garage, because the garage blocks external signals, the scene recognition device may switch to one or more different cells, or even lose signal connection. This reduction or disappearance of cell count can serve as one of the important criteria for determining whether an entry into an underground parking garage has occurred.

[0089] In some embodiments, cell signal strength change information refers to the trend of change in the signal strength of the currently serving cell received by the scene recognition device. Typically, signal strength decreases significantly upon entering a parking garage, and gradually increases upon leaving the garage due to stronger base station signals outside. Therefore, the increasing signal strength in cell signal strength change information can serve as an important indicator for determining whether the scene is leaving a parking garage. Monitoring this trend can help determine whether the scene is in a parking garage environment.

[0090] In some embodiments, combining real-time status information with cell count change information and / or cell signal strength change information can more accurately identify whether a scene is being left behind in the parking garage. This utilizes the combination of motion status information and cell count change information, avoiding potential misjudgments from single sensor data and improving the accuracy of scene recognition.

[0091] In some embodiments, by combining real-time status information detected by the inertial measurement unit (IMU) as an uphill state with cell number change information as an increasing trend, it is possible to effectively identify whether a scenario is being exited from the underground parking garage. This method can improve the robustness and accuracy of scene recognition, thereby enhancing the timing and effectiveness of subsequent network optimization strategies, and ultimately improving the communication experience when switching between inside and outside the underground parking garage.

[0092] In some embodiments, combining real-time status information with cell signal strength change information can further verify whether the scene is an exit from the underground parking garage. For example, if the real-time status information indicates an uphill state, and the cell signal strength change information simultaneously shows an increase in signal strength, it can be inferred that the scene is an exit from the underground parking garage. This approach utilizes not only motion status information but also wireless signal characteristics, enhancing the reliability of scene recognition.

[0093] For example, in some embodiments, by combining real-time status information detected by the inertial measurement unit (IMU) as an uphill state with cell number change information as an increase in number, and / or cell signal strength change information as a trend of increasing signal strength, it is possible to effectively identify whether the scene is leaving the underground parking garage.

[0094] In some embodiments, combining real-time status information with cell count change information and / or cell signal strength change information can more accurately identify whether a scene is entering an underground parking garage. This utilizes the combination of motion status information and cell count change information, avoiding potential misjudgments from single sensor data and improving the accuracy of scene recognition.

[0095] In some embodiments, by combining real-time status information detected by the inertial measurement unit (IMU) as a downhill state with cell number change information as a decreasing trend, it is possible to effectively identify whether a scenario is entering an underground parking garage. This method can improve the robustness and accuracy of scene recognition, thereby enhancing the timing and effectiveness of subsequent network optimization strategies, and ultimately improving the communication experience when switching between inside and outside the underground parking garage.

[0096] In some embodiments, combining real-time status information with cell signal strength change information can further verify whether the scene is an entry into an underground parking garage. For example, if the real-time status information indicates a downhill state, and the cell signal strength change information simultaneously indicates a weakening signal, it can be inferred that the scene is an entry into an underground parking garage. This approach utilizes not only motion status information but also wireless signal characteristics, enhancing the reliability of scene recognition.

[0097] For example, in some embodiments, by combining real-time status information detected by the inertial measurement unit (IMU) as downhill status with cell number change information as decreasing number, and / or cell signal strength change information as a weakening trend of signal strength, it is possible to effectively identify whether the scene is entering the underground parking garage.

[0098] In some embodiments, the scene recognition result is a final judgment derived from one or more of the following: real-time status information, cell number change information, and cell signal strength change information. The scene recognition result can be represented as entering the underground parking garage, leaving the underground parking garage, or still being in the underground parking garage. The scene recognition result will be used for subsequent network policy adjustments, content preloading, or other intelligent service decisions.

[0099] In the embodiments of this application, by determining the scene recognition result based on one or more of the real-time status information, cell number change information, and cell signal strength change information, multi-dimensional cross-validation can be achieved, thereby improving the robustness and accuracy of underground parking scene recognition, and thus better supporting network optimization and intent prediction, and improving the overall user experience.

[0100] Therefore, in the embodiments of this application, efficient identification of underground parking garage scenes is achieved by integrating multiple information sources, including one or more of sensor information, cell number change information, and cell signal strength change information. Among these, cell number change information and / or cell signal strength change information, as well as other types of cell information detected, can be used to assist in determining whether one has entered or left the underground parking garage.

[0101] For example, in some embodiments, the real-time status of the scene recognition device is first obtained through sensor information (real-time status information), and then combined with changes in cell information at the network layer (changes in the number of cells and / or changes in cell signal strength) to jointly determine whether it is in a parking garage scene, thereby providing a reliable basis for subsequent network optimization and service decisions.

[0102] In the embodiments of this application, when scene recognition of the underground parking scene is performed based on the first real-time information and the scene recognition result is obtained, if the satellite search information indicates that no satellite was found, the duration of the period during which no satellite was found is determined; if the duration is greater than a preset time threshold, the scene recognition result is determined to be an underground parking scene.

[0103] In the embodiments of this application, when scene recognition of a parking garage scene is performed based on the first real-time information and the scene recognition result is obtained, if the satellite search information indicates that a satellite has been found, the real-time moving speed is determined; if the real-time moving speed is greater than a preset speed threshold, the scene recognition result is determined to be a non-parking garage scene.

[0104] In some embodiments, satellite search information refers to signal status information acquired by a scene recognition device (such as a smartphone) when attempting to connect to the Global Positioning System (GPS). Satellite search information is used to determine whether at least a certain number of satellite signals can be received to achieve positioning functionality. When the scene recognition device cannot detect any available satellite signals, it indicates a state where no satellites have been found.

[0105] In some embodiments, the duration of no satellite detection refers to the length of time during which the device continuously fails to receive a valid satellite signal. The duration of no satellite detection is an important basis for determining whether to enter an area with no satellite signal coverage, such as an underground parking garage. For example, in environments such as urban underground parking lots, subway stations, and tunnels, GPS signals are often blocked by buildings, causing devices to be unable to receive satellite signals for extended periods.

[0106] In some embodiments, by monitoring the duration of no satellite detection, a preliminary determination can be made as to whether an underground parking garage scene has been entered. This method, which monitors the duration of no satellite detection, has the advantages of low power consumption and no reliance on third-party services, making it suitable for various mobile terminal devices. Simultaneously, it avoids dependence on high-precision sensors, thereby reducing power consumption and improving the system's versatility, thus enabling wider application in scene recognition in various underground environments.

[0107] In some embodiments, a preset time threshold can be used for scene recognition. The preset time threshold can be any value greater than 0, and this application does not impose specific limitations on it.

[0108] For example, in some embodiments, the preset time threshold can be a time standard set based on actual testing and experience, used to distinguish between normal outdoor environments and indoor environments such as basements without satellite signal coverage. For instance, if the scene recognition device occasionally and briefly loses satellite signal in an outdoor environment, this may be due to tree or building obstruction and will not be considered as entering a basement; however, if the scene recognition device fails to receive satellite signal for more than the preset time threshold (e.g., 30 seconds to 1 minute), it can be reasonably inferred that it has entered a basement scene.

[0109] In some embodiments, by comparing the duration of no satellite detection with a preset time threshold, it is possible to accurately determine whether a parking garage scene has been entered, improving the reliability of scene recognition and thus reducing the false positive rate. Furthermore, since it only requires monitoring the satellite signal status and does not involve complex calculations or additional hardware, it offers good energy efficiency and practicality.

[0110] In other words, in the embodiments of this application, by monitoring the duration of no satellite search and comparing it with a preset time threshold, the underground parking scenario can be identified efficiently and with low power consumption. By monitoring the duration of no satellite search and comparing it with a preset time threshold, reliance on third-party services can be avoided, thereby improving the system's autonomy and flexibility.

[0111] In some embodiments, when the satellite search information indicates that satellites have been found, for example, when the scene recognition device receives a GPS signal and successfully searches for at least a certain number of satellites, the real-time movement speed can be further determined.

[0112] In some embodiments, the real-time moving speed of the scene recognition device can be derived from the motion trajectory data calculated by the receiver. The real-time moving speed is typically expressed in meters per second (m / s) or kilometers per hour (km / h) and is used to determine whether the device is in a rapid moving state.

[0113] In some embodiments, obtaining real-time movement speed can help determine the type of scene. For example, if the real-time movement speed is greater than a preset speed threshold, the scene identification result can be considered as a non-parking scene.

[0114] In some embodiments, a preset speed threshold can be used for scene recognition. The preset speed threshold can be any value greater than 0, and this application does not impose specific limitations on it.

[0115] In some embodiments, the preset speed threshold can be a value set according to the actual application scenario to determine whether the vehicle is in a relatively fast-moving state (such as driving). For example, the preset speed threshold can be set to 5 km / h, and if the real-time moving speed exceeds this value, it is considered to be in a non-parking scenario.

[0116] In some embodiments, movement within a basement is mostly slow, while movement speed in outdoor or open areas is typically higher. Therefore, by combining the presence or absence of a GPS signal with the magnitude of the real-time movement speed, when the satellite search information indicates that a satellite has been found and the real-time movement speed exceeds a preset speed threshold, the current scenario can be determined to be a non-basement scenario.

[0117] In some embodiments, upon receiving satellite signals, the real-time movement speed is further determined and compared with a preset speed threshold, thereby achieving accurate identification of the underground parking garage scene. That is, by combining satellite signal detection with real-time movement speed analysis, efficient identification of the underground parking garage scene is achieved.

[0118] For example, in some embodiments, the condition for further determining the real-time moving speed is first determined by detecting whether a sufficient number of satellite signals are received; if the condition is met, the real-time moving speed is calculated and compared with a preset speed threshold, and finally the scene recognition result is output.

[0119] In the embodiments of this application, when scene recognition of the underground parking scene is performed based on the first real-time information and the scene recognition result is obtained, if the cell monitoring result is that a preset fence cell is detected, the scene recognition result is determined to be an underground parking scene; if the cell monitoring result is that a preset fence cell is not detected, the scene recognition result is determined to be a non-underground parking scene.

[0120] In some embodiments, cell monitoring results may include current serving cell information obtained by a scene recognition device through monitoring cellular network signals, which can be used to determine whether the location is in a specific geographical area. The monitoring process is typically performed by the mobile communication module, requiring no additional hardware support and featuring low power consumption and high efficiency.

[0121] In some embodiments, the preset fence cells associated with basement scene recognition can be a set of cells pre-defined by the scene recognition device, which cover a known basement entrance or interior area.

[0122] For example, in some embodiments, the construction of preset fenced cells can be derived from historical data statistics, such as recording the cells passed through when frequently entering and exiting the underground parking garage, and using the cells with the highest frequency or longest duration as fenced cells.

[0123] In some embodiments, determining whether an entry into an underground parking garage can be quickly achieved by checking whether a pre-defined fenced cell has been detected through cell monitoring. This method of identifying underground parking garage scenes based on cell monitoring results and pre-defined fenced cells does not rely on GPS or IMU sensors, making it suitable for situations with no satellite signal or significant changes in phone orientation. Furthermore, it reduces reliance on third-party services, thereby enhancing the system's autonomy and flexibility.

[0124] For example, in some embodiments, when a preset fence cell is detected, it can be determined that the scene recognition device has entered the preset fence cell, and at this time it can be determined that the scene recognition device may be in a parking garage environment. When no preset fence cell is detected, it indicates that the current location is not within the known parking garage area, and therefore it can be determined to be a non-parking garage scene.

[0125] Therefore, in the embodiments of this application, by monitoring cellular information and combining it with a preset fenced cell judgment mechanism, efficient identification of underground parking scenarios can be achieved. Furthermore, identifying underground parking scenarios based on cell monitoring results and preset fenced cells reduces reliance on external sensors and third-party services, thereby enhancing the system's autonomy and adaptability. This allows for wide application in various underground parking environments, significantly improving the communication and service experience within underground parking garages.

[0126] In the embodiments of this application, in the case of an underground parking garage scenario, the cells that can be monitored are recorded; and a preset fenced cell is determined based on the cells that can be monitored.

[0127] In some embodiments, the scene recognition device may pre-build a preset fenced area.

[0128] For example, in some embodiments, when constructing a preset fence cell, if the location is within an underground parking garage, a cellular network information collection mechanism can be activated to scan and record all currently detectable cells. This detectable cell information includes the serving cell and neighbor cells, and contains parameters such as the frequency, PCI (Physical Cell Identifier), and RSRP (Reference Signal Received Power) of each detectable cell. By recording this information, a cellular network environment map of the current underground parking garage can be constructed, providing data support for subsequent electronic fence construction.

[0129] In some embodiments, the process of recording the cells that can be monitored is typically performed by the wireless communication module of the scene recognition device, which has the ability to periodically scan the surrounding cells that can be monitored. After entering the underground parking garage, the number and strength of cellular signals received by the scene recognition device will change due to the obstruction of the garage structure. Recording all the information of the cells that can be monitored at this time helps to determine whether the person remains in the same underground parking garage or has left the garage area.

[0130] In some embodiments, the scene recognition device can perform statistical analysis based on recorded listenable cell information, such as calculating the frequency or duration of a particular listenable cell, to determine the listenable cells most likely representing the characteristics of the parking garage. These listenable cells will be used to construct electronic fences, enabling more accurate parking garage scene recognition.

[0131] In the embodiments of this application, after determining the scene recognition result for the underground parking garage scene based on the first real-time information, the scene recognition device can further trigger corresponding processing logic. For example, if the scene recognition result is an underground parking garage scene, the scene recognition device can choose to switch network strategies, optimize signal connections, etc.

[0132] In summary, the scene recognition method proposed in this application integrates multiple types of real-time information (such as sensor information, cellular network information, satellite information, etc.) to comprehensively determine whether a scene is in a parking garage, thereby achieving multimodal and high-precision scene recognition. Compared with existing technologies that rely on a single sensor or third-party services, this solution improves the accuracy and universality of recognition while reducing dependence on third-party data.

[0133] This application proposes a scene recognition method to acquire first real-time information. The first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; and scene recognition of a parking garage scene based on the first real-time information to obtain a scene recognition result. Therefore, in the embodiments of this application, one or more types of real-time information related to the scene recognition device itself can be collected, such as sensor information, cell number change information, cell signal strength change information, satellite search information, etc. By fusing multiple types of real-time information, a comprehensive judgment and full recognition of the parking garage scene can be performed, thereby achieving multimodal, high-precision scene recognition and improving the accuracy and universality of scene recognition.

[0134] Based on the above embodiments, another embodiment of this application proposes a scene recognition method. In view of the problem that the data provided by third-party applications in related technologies are limited in spatial coverage or have large errors in the calculation of attitude angles, resulting in inaccurate scene recognition, this method integrates multiple types of real-time information (such as sensor information, cellular network information, satellite information, etc.) to comprehensively determine whether the scene is in a parking lot, thereby achieving multimodal and high-precision scene recognition.

[0135] The following describes the scene recognition method proposed in this application by taking a mobile device as an example.

[0136] The scene recognition method proposed in this application can achieve scene recognition using data collected by an IMU. Specifically, the data collected by the IMU can be used to identify scenes of entering and exiting a parking garage via a ramp.

[0137] Exemplarily, in some embodiments, such as Figure 2 As shown, the identification of uphill and downhill slopes using data acquired by the IMU can be calculated based on acceleration and gyroscope data. First, data is acquired using the accelerometer and gyroscope (step 201), then it is determined whether the system is in a quasi-static state (step 202). If not, data acquisition and processing continue; if so, the attitude array is calculated, and the acquired sensor data is converted to a horizontal coordinate system (assuming a vehicle frame), and the pitch angle is calculated (step 203). Further, it is checked whether the absolute value of the pitch angle (absolute value of the pitch angle sequence) satisfies the uphill / downhill state (step 204). If not, data acquisition and processing continue; if so, the uphill / downhill state is further determined and output (step 205).

[0138] Exemplarily, in some embodiments, such as Figure 3As shown, when determining the uphill / downhill state (real-time state information), the cached data can be initialized first (step 301), including but not limited to initializing the following data: quasi-static flag StaticFlag (initFlag); valid value for identifying the state validFlag; potential uphill / downhill state hillDetectStateTemp; a new queue imuValue; and a cached queue pitch_smooth after pitch smoothing. Next, the sensor can be started to collect sensor data (step 302), and the sampling frequency can be set (default 15Hz). The collected data includes, but is not limited to, accelerometer (ACC) data (acceleration data, such as acceleration change values) and gyroscope (gyroscope data). Then, the collected sensor data can be stored (step 303), for example, by writing the ACC data and gyroscope data into the raw data queue imuValue, with a default queue size of 15. Next, it can be determined whether it is quasi-static based on the sensor data, and the quasi-static flag StaticFlag is updated (step 304). Then, based on the quasi-static determination result (the value of the quasi-static flag StaticFlag), the subsequent uphill and downhill states can be determined (step 305).

[0139] Exemplarily, in some embodiments, such as Figure 4As shown, in the process of determining the uphill / downhill state based on StaticFlag, if the quasi-static flag StaticFlag indicates that it is not quasi-static, then we can first check whether the queue size imuValue is 15 (step 401). If not, we continue to perform data collection and processing; if so, we perform data calculation (step 402), for example, we calculate the mean of the three axes gyro and form a matrix [g1, g2, g3], and obtain the modulus g_norm. At the same time, we calculate the mean of the three axes acc and form a matrix [a1, a2, a3]. Then, we perform the determination of the principal axis based on the data calculation results. First, we perform the first judgment, that is, we judge whether the first condition for the existence of a principal axis is met (step 403). For example, we judge whether Abs(g_norm) is less than or equal to the gyro static threshold quasiStaticThreshold (default 0.02), and whether a3 is less than the static principal axis control threshold (default 7). If the condition is not met, the queue imuValue is cleared (step 404). If the condition is met, the data calculation continues to obtain matrix C3 (step 405). For example, the modulus a_norm is obtained first based on [a1, a2, a3]. [a1, a2, a3] is divided by a_norm to obtain a new matrix acc_normlized, and C3 = acc_normlized. It is determined whether abs(C3[0]) > 0.5 (step 406). If it is met, C2 = [C3[1], -C3[0], 0] (step 407). If it is not met, C2 = [C3[1], -C3[0], 0] (step 408). Then the attitude matrix Cbh is calculated (step 409). For example, C2 = C2 / C2 modulus, C1 = C2 cross product C3, and the attitude matrix Cbh = a new matrix of C1, C2, and C3 stacked row by row. Finally, subsequent processing is performed based on the attitude matrix (step 410).

[0140] Exemplarily, in some embodiments, such as Figure 5 As shown, after calculating the attitude matrix Cbh, the following processes can be performed sequentially:

[0141] Step 501: Determine whether acc_normlized[0] > mainAxisTh, where mainAxisTh defaults to 0.7. If it is satisfied, proceed to step 502; otherwise, proceed to step 503.

[0142] Step 502: Set the process variable Ctemp = [[0, 1, 0], [-1, 0, 0], [0, 0, 1]], and set the main axis mainAxis to landscape mode (make sure the main axis is landscape).

[0143] Step 503: Determine if acc_normlized[0] < mainAxisTh. If it is satisfied, proceed to step 504; otherwise, proceed to step 505.

[0144] Step 504: Set the process variable Ctemp = [[0, -1, 0], [1, 0, 0], [0, 0, 1]], and set the main axis mainAxis to landscape mode (make sure the main axis is landscape).

[0145] Step 505: Determine if abs(acc_normlized[1])>mainAxisTh is satisfied. If satisfied, proceed to step 506; otherwise, proceed to step 507.

[0146] Step 506: Set the process variable Ctemp = [[1, 0, 0], [0, 1, 0], [0, 0, 1]], and set the main axis mainAxis to portrait mode (make sure the main axis is vertical).

[0147] Step 507: Determine that the main axis is not found, and clear the imuValue.

[0148] Step 508: Update parameters, for example, update Cbh based on the result of multiplying Ctemp and Cbh; set staticFlag to true; initialize the attitude quadruple q = [1, 0, 0, 0]; clear imuValue.

[0149] Exemplarily, in some embodiments, such as Figure 6As shown, in the process of determining the uphill / downhill state based on StaticFlag, if the quasi-static flag StaticFlag indicates quasi-static, then the parameters AccTrans and AccTrans can be calculated first (step 601). Here, AccTrans = acc × Cbh_t, GyroTrans = gyro × Cbh_t, and Cbh_t is the transpose of the attitude matrix Cbh. Next, the attitude quadruple q is updated according to AccTrans and AccTrans (step 602). For example, a new attitude quadruple q can be calculated based on the initial attitude quadruple q, accTrans, gyroTrans, and the time interval deltaTs between the current and previous acquisitions of acc. Next, the attitude quadruple q is converted into attitude angle att using the quaternion-to-attitude-angle conversion method (step 603). Then, the pitch angle cache queue pitch_record is updated according to the attitude angle att, and the mean pitch angle meanTemp is calculated (step 604). For example, (att[0]×180 / π) is stored in the pitch angle buffer queue pitch_record. The size of pitch_record can be set according to int(sampling frequency×1.5). The pitch angle meanTemp can be determined according to the mean of pitch_record. Then, the pitch angle meanTemp is stored in the pitch angle smoothed buffer queue pitch_smooth (step 605).

[0150] Exemplarily, in some embodiments, such as Figure 7 As shown, after storing the pitch angle meanTemp into the pitch angle smoothed cache queue pitch_smooth (i.e., step 605), we can first check if the queue imuValue size is 15 (step 701). If not, we proceed to step 710; if so, we perform data calculation (step 702), for example, calculating the mean of the three gyro axes and forming matrices [g1, g2, g3], and obtaining the modulus g_norm. At the same time, we calculate the mean of the three acc axes and form matrices [a1, a2, a3]. Next, we check if Abs(g_norm) is less than or equal to the gyro static threshold quasiStaticThreshold (default 0.02) (step 703). If not, we clear the queue imuValue (step 704); if so, we continue data calculation to obtain matrix C3 (step 705). For example, first obtain the modulus a_norm from [a1, a2, a3], then divide [a1, a2, a3] by a_norm to obtain a new matrix acc_normlized, C3 = acc_normlized. Then, the following processing can be performed sequentially:

[0151] Step 706: Determine if mainAxis = landscape and Acc_normalized[1] > mainAxisTh? If it is satisfied, proceed to step 707; otherwise, proceed to step 708.

[0152] Step 707: Set the main axis to portrait mode (make sure the main axis is vertical), and update staticFlag, i.e., set staticFlag = false.

[0153] Step 708: Determine if mainAxis = vertical screen and abs(Acc_normalized[0]) > mainAxisTh? If it is satisfied, proceed to step 709; otherwise, proceed to step 704.

[0154] Step 709: Set the main axis to landscape mode (make sure the main axis is horizontal), and update staticFlag, i.e., set staticFlag = false.

[0155] Step 710: Determine if the size of pitch_smooth has reached the threshold, for example, is it the full threshold of 15? If so, proceed to step 711; otherwise, continue building the pitch-smooth buffer queue.

[0156] Step 711: Determine medVal based on pitch_smooth, and determine the uphill / downhill state based on medVal. For example, determine medVal based on the median of pitch_smooth.

[0157] Exemplarily, in some embodiments, such as Figure 8 As shown, when determining the uphill / downhill state based on medVal, the following processing can be further performed:

[0158] Step 801: Determine if hillDetectStateTemp = uphill. If yes, proceed to step 802; otherwise, proceed to step 803.

[0159] Step 802: Determine if medVal > uphill threshold. The uphill threshold can be uphillTh, with a value of 3.5. If yes, proceed to step 804; otherwise, proceed to step 805.

[0160] Step 803: Determine if hillDetectStateTemp = downhill. If yes, proceed to step 806; otherwise, proceed to step 807.

[0161] Step 804: Determine if the conditions for continuous uphill are met. For example, determine the uphill time using the current timestamp and the uphill start timestamp upHillStartTs, and then determine if the uphill time is greater than the duration periodTh 2 and the uphill start timestamp upHillStartTs > 0. If yes, proceed to step 808; otherwise, proceed to step 809.

[0162] Step 805: Set upHillStartTs=0 and hillDetectStateTemp=0.

[0163] Step 806: Determine if medVal < downhill threshold. The downhill threshold can be downhillTh, with a value of -3.5. If yes, proceed to step 810; otherwise, proceed to step 811.

[0164] Step 807: Report non-uphill / downhill status.

[0165] Step 808: Set upHillStartTs=0 and report the uphill status.

[0166] Step 809: Determine the potential state.

[0167] Step 810: Determine if the conditions for a continuous downhill slope are met. For example, determine the downhill time using the current timestamp and the downhill start timestamp downHillStartTs, and then determine if the downhill time is greater than the duration periodTh 2 and the downhill start timestamp downHillStartTs > 0. If yes, proceed to step 812; otherwise, proceed to step 809.

[0168] Step 811: Set downHillStartTs=0 and hillDetectStateTemp=0.

[0169] Exemplarily, in some embodiments, such as Figure 9 As shown, when determining the potential state, the following processing can be further performed:

[0170] Step 901: Determine if medVal > uphill and hillDetectStateTemp = 0. If so, proceed to step 902; otherwise, proceed to step 903.

[0171] Step 902: Set the current time as the start time of the uphill climb, and set hillDetectStateTemp = uphill.

[0172] Step 903: Determine whether medVal < downhill and hillDetectStateTemp = 0 are satisfied? If satisfied, execute Step 904.

[0173] Step 904: Set the current time as the start time of downhill, and at the same time set hillDetectStateTemp = downhill.

[0174] Step 905: Clear the cache queue pitch_smooth after smoothing the pitch angle.

[0175] In an embodiment of the present application, after identifying the uphill and downhill states (real-time state information) from the data collected by the IMU, the cellular network signal strength can be further combined to identify the basement scenario.

[0176] Exemplarily, in some embodiments, as Figure 10 shown, it can be first determined whether the downhill state and the number of cellular cells decreasing are satisfied (Step 1001), that is, whether the real-time state information is in the downhill state and the cell quantity change information is a decrease in quantity. If satisfied, the state of entering the basement can be reported, that is, the scenario recognition result is determined to be the scenario of entering the basement (Step 1002); otherwise, it is determined whether the downhill state and the cellular cell signal becoming weaker are satisfied (Step 1003), that is, whether the real-time state information is in the downhill state and the cell signal strength change information is a decrease in signal strength. If satisfied, the scenario recognition result can be reported as the scenario of entering the basement (Step 1002); otherwise, it can be determined as a non-basement scenario (Step 1004).

[0177] That is to say, the judgment logic for the scenario of entering the basement is to identify the IMU downhill event, and at the same time identify the cellular count decreasing event and / or the cellular serving cell signal becoming weaker, then the state of entering the basement can be determined.

[0178] Exemplarily, in some embodiments, as Figure 11 shown, it can be first determined whether the uphill state and the number of cellular cells increasing are satisfied (Step 1101), that is, whether the real-time state information is in the uphill state and the cell quantity change information is an increase in quantity. If satisfied, the state of leaving the basement can be reported, that is, the scenario recognition result is determined to be the scenario of leaving the basement (Step 1102); otherwise, it is determined whether the uphill state and the cellular cell signal strengthening are satisfied (Step 1103), that is, whether the real-time state information is in the uphill state and the cell signal strength change information is an increase in signal strength. If satisfied, the scenario recognition result can be reported as the scenario of leaving the basement (Step 1102); otherwise, it can be determined as a non-basement scenario (Step 1104).

[0179] In other words, the logic for determining the exit from the underground parking garage scenario is to identify the IMU uphill event, and at the same time identify the cellular count increment event and / or the cellular serving cell signal becoming stronger, thus determining the exit from the underground parking garage status.

[0180] The scene recognition method proposed in this application can achieve scene recognition using data from a GPS module. For example, Figure 12 As shown, one can determine whether a mobile terminal is inside a basement by checking whether it can receive satellite signals.

[0181] For example, in some embodiments, if a satellite signal can be received and the movement speed is greater than the walking speed of a person, it is considered to be outside the underground parking lot, that is, the scene recognition result is an underground parking lot scene; if no satellite is found for a period of time, it is considered to be inside the underground parking lot, that is, the scene recognition result is a non-underground parking lot scene.

[0182] Exemplarily, in some embodiments, such as Figure 13 As shown, when navigation software initiates a GPS request, a judgment is made. If the mobile terminal can receive satellite signals and the movement speed is greater than the walking speed of a person, it is considered to be outside the underground parking lot; otherwise, if no satellites are found for a period of time, it is considered to be inside the underground parking lot. Specifically, it can first determine if an application actively requests GPS positioning (step 1301). If so, it further determines whether the number of satellites received is not 0 (step 1302). If so, the current movement speed can be calculated based on GPS (step 1303). If the current movement speed is greater than a threshold (e.g., 1 m / s), it is determined to be outside the underground parking lot, i.e., the scene identification result is determined to be a non-underground parking lot scene (step 1305). If the number of satellites received is 0, it further determines whether no satellites are found for a period of time (step 1304). If so, it can be determined to be inside the underground parking lot, i.e., the scene identification result is determined to be an underground parking lot scene (step 1306).

[0183] The scene recognition method proposed in this application can achieve scene recognition through the setting of electronic fences. This includes fence construction (construction of preset fence zones) and fence recognition (scene recognition based on preset fence zones).

[0184] Exemplarily, in some embodiments, such as Figure 14As shown, during the construction of the preset fenced cells, while reporting events, the corresponding cellular network information is recorded. The cell that appears most frequently is designated as the fenced cell (preset fenced cell) and stored in the database. For example, after reporting into the underground parking garage, cell information is recorded for a period of time after entering the garage, and the cell with the longest duration or the most frequent occurrence is designated as the fenced cell (step 1401). After reporting out of the underground parking garage, cell information is recorded for a period of time before entering the garage, and the cell with the longest duration or the most frequent occurrence is designated as the fenced cell (step 1402). Finally, the constructed preset fenced cells can be stored in the local database (step 1403).

[0185] Exemplarily, in some embodiments, such as Figure 15 As shown, when performing scene recognition based on a preset fenced cell, the system determines whether to enter the preset fenced cell based on the obtained cell monitoring results (cellular cell monitoring information) (step 1501). If so, it reports entering the fence (equivalent to underground parking), and the scene recognition result can be determined to be an underground parking scene (step 1502); otherwise, the scene recognition result can be determined to be a non-underground parking scene (step 1503).

[0186] Exemplarily, in some embodiments, such as Figure 16 As shown, when performing scene recognition based on a preset fence cell, the system determines whether to enter the preset fence cell based on the obtained cell monitoring results (cellular cell monitoring information) (step 1601). If so, it reports entering the fence (equivalent to the underground parking lot), and the scene recognition result can be determined as entering the underground parking lot scene (step 1602).

[0187] Exemplarily, in some embodiments, such as Figure 17 As shown, when performing scene recognition based on a preset fence cell, the system determines whether to exit the preset fence cell based on the obtained cell monitoring results (cellular cell monitoring information) (step 1701). If so, it reports exiting the fence (equivalent to the underground parking lot), and the scene recognition result can be determined as leaving the underground parking lot scene (step 1702).

[0188] In the embodiments of this application, fence identification can be used when both IMU identification and GPS identification are not satisfied, and only cellular information is monitored, achieving zero power consumption.

[0189] In summary, compared with related technologies that rely on a single sensor or third-party services, the scene recognition method proposed in this application improves the accuracy and universality of recognition while reducing reliance on third-party data.

[0190] The scene recognition method proposed in this application has the following advantages:

[0191] 1. Highly versatile and independent of third parties: The solution itself can recognize underground parking scenes without relying on third-party recognition solutions, ensuring autonomy and flexibility.

[0192] 2. Coverage of a larger number of parking garages: Compared with other related solutions, this solution can identify and cover a significantly larger number of underground parking garages.

[0193] 3. Precisely Identify and Address User Pain Points: Currently, over 80% of users encounter problems such as slow internet speeds, poor signal, and difficulty scanning QR codes in underground parking garages. This solution provides precise scene identification capabilities, including cellular network optimization, offering key scene information to help improve network signal strength and service quality in underground parking garages. It also includes user intent prediction, providing accurate scene input to make predicting user needs in underground parking garages (such as downloading content, preloading pages, switching network strategies, etc.) more reliable.

[0194] 4. Comprehensive improvement of user experience: By providing effective support for network optimization and intent prediction, it ultimately significantly improves the user experience in all aspects of communication, networking, and service access within the parking garage.

[0195] In summary, the scene recognition method proposed in this application has the following advantages: First, it enables multimodal scene recognition, accurately identifying user entry and exit scenarios in underground parking garages through various technologies (including IMU, GPS, and electronic fences). Second, it achieves universal scene coverage, universally recognizing various underground parking garage scenarios and is applicable to garages of different types and structures. Third, it achieves broad garage coverage, covering a wide number of garages and scalable to a large number of underground parking environments. Fourth, it improves user adaptability, applicable to users entering and exiting underground parking garages in various ways (such as walking, driving, etc.), enhancing the user experience. Fifth, it enables self-learning and data expansion, possessing self-learning capabilities to continuously expand and optimize underground parking garage data without manual intervention. Sixth, it achieves low-power design, taking into account the low power consumption of mobile phones, ensuring efficient system operation on mobile devices and extending battery life.

[0196] This application proposes a scene recognition method that can collect one or more types of real-time information related to the scene recognition device itself, such as sensor information, cell number change information, cell signal strength change information, satellite search information, and other types of real-time information. By fusing multiple types of real-time information, a comprehensive judgment and full recognition of the underground scene is made, thereby achieving multimodal and high-precision scene recognition, which can improve the accuracy and universality of scene recognition.

[0197] Based on the above embodiments, in another embodiment of this application... Figure 18 This is a schematic diagram of the composition structure of the scene recognition device proposed in the embodiments of this application, as shown below. Figure 18As shown, the scene recognition device 110 proposed in this application embodiment may include:

[0198] The acquisition unit 1101 is used to acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results;

[0199] The recognition unit 1102 is used to perform scene recognition of the underground parking scene based on the first real-time information and obtain the scene recognition result.

[0200] In the embodiments of this application, further, Figure 19 This is a schematic diagram of the composition structure of the electronic device proposed in the embodiments of this application, such as... Figure 19 As shown, the electronic device 120 proposed in this application embodiment may include a processor 1201, a memory 1202, a communication interface 1203, and a bus 1204 for connecting the processor 1201, the memory 1202 and the communication interface 1203.

[0201] In the embodiments of this application, the processor 1201 can be at least one of the following: Application-Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field-Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that for different devices, the electronic device used to implement the above-mentioned processor function can also be other types, and this application embodiment does not specifically limit this. The electronic device 120 may also include a memory 1202, which can be connected to the processor 1201. The memory 1202 is used to store executable program code, which includes computer operation instructions. The memory 1202 may include high-speed RAM memory and may also include non-volatile memory, such as at least two disk drives.

[0202] In embodiments of this application, bus 1204 is used to connect communication interface 1203, processor 1201, and memory 1202, as well as the mutual communication between these devices.

[0203] In practical applications, the aforementioned memory 1202 can be volatile memory, such as random-access memory (RAM); or non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); or a combination of the above types of memory, and provide instructions and data to the processor 1201.

[0204] Furthermore, in an embodiment of this application, the processor 1201 is configured to: acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; and perform scene recognition of the underground parking lot scene based on the first real-time information to obtain scene recognition results.

[0205] Furthermore, in this embodiment, the functional modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0206] If the integrated unit is implemented as a software functional module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to related technologies, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method of this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0207] This application provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements the scene recognition method described above.

[0208] Specifically, the program instructions corresponding to a scene recognition method in this embodiment can be stored on storage media such as optical discs, hard disks, and USB flash drives. When the program instructions corresponding to a scene recognition method in the storage media are read or executed by an electronic device, the following steps are included:

[0209] Acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results;

[0210] Scene recognition is performed on the underground parking lot scene based on the first real-time information to obtain the scene recognition results.

[0211] This application also provides a computer program product.

[0212] In some embodiments, the computer program product may include a computer program or instructions.

[0213] In some embodiments, the computer program product can be applied to the computer device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the computer device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0214] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0215] This application is described with reference to schematic and / or block diagrams of implementations of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the schematic and / or block diagrams can be implemented by computer program instructions, and combinations of blocks in the schematic and / or block diagrams can be implemented. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the schematic and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0216] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the implementation flow diagram. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0217] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0218] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. A scene recognition method, characterized in that, The method includes: Acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; Based on the first real-time information, scene recognition of the underground parking lot scene is performed to obtain scene recognition results.

2. The method according to claim 1, characterized in that, The process of scene recognition based on the first real-time information to obtain scene recognition results includes: Real-time status information is determined based on the sensor information; Based on the real-time status information, the cell number change information, and / or cell signal strength change information, the scene recognition result is determined.

3. The method according to claim 2, characterized in that, The determination of the scene recognition result based on the real-time status information, the cell number change information, and / or cell signal strength change information includes: If the real-time status information indicates an uphill state and the cell number change information indicates an increase in the number of cells, then the scene recognition result is determined to be a scene of leaving the underground parking garage; and / or, If the real-time status information indicates an uphill state and the cell signal strength change information indicates an increase in signal strength, then the scene recognition result is determined to be a scene of leaving the underground parking lot.

4. The method according to claim 2, characterized in that, The determination of the scene recognition result based on the real-time status information, the cell number change information, and / or cell signal strength change information includes: If the real-time status information indicates a downhill state and the cell number change information indicates a decrease in the number of cells, then the scene recognition result is determined to be an entry into the underground parking garage; and / or, If the real-time status information indicates a downhill state and the cell signal strength change information indicates a weakening signal strength, then the scene recognition result is determined to be a scene of entering the underground parking garage.

5. The method according to any one of claims 1 to 4, characterized in that, The sensor information includes acceleration data and gyroscope data, and the acquisition of the first real-time information includes: The acceleration data is acquired using an accelerometer. The gyroscope data is obtained through the gyroscope.

6. The method according to claim 5, characterized in that, The process of determining real-time status information based on the sensor information includes: Based on the acceleration data and / or the gyroscope data, determine the real-time dynamic and static state; When the real-time dynamic and static state is non-static, the real-time pitch angle is determined based on the acceleration data and / or the gyroscope data; The real-time status information is determined based on the real-time pitch angle.

7. The method according to claim 1, characterized in that, The process of scene recognition based on the first real-time information to obtain scene recognition results includes: If the satellite search information indicates that no satellite was found, determine the duration of the period during which no satellite was found; If the duration exceeds a preset time threshold, the scene recognition result is determined to be a parking garage scene.

8. The method according to claim 7, characterized in that, The process of scene recognition based on the first real-time information to obtain scene recognition results includes: If the satellite search information indicates that a satellite has been found, determine the real-time movement speed; If the real-time moving speed is greater than a preset speed threshold, the scene recognition result is determined to be a non-parking scene.

9. The method according to claim 1, characterized in that, The process of scene recognition based on the first real-time information to obtain scene recognition results includes: If the cell monitoring result indicates that a preset fenced cell has been detected, the scene recognition result is determined to be a basement scene. If the cell monitoring result indicates that no preset fenced cell has been detected, the scene identification result is determined to be a non-parking garage scene.

10. The method according to claim 9, characterized in that, The method further includes: In a basement setting, record the cells that can be monitored; The preset fenced cell is determined based on the cells that can be monitored.

11. A scene recognition device, characterized in that, The scene recognition device includes: The acquisition unit is configured to acquire first real-time information; wherein the first real-time information includes at least one or more of the following: sensor information; cell number change information; cell signal strength change information; satellite search information; cell monitoring results; The recognition unit is used to perform scene recognition of the underground parking lot scene based on the first real-time information and obtain the scene recognition result.

12. An electronic device, characterized in that, The electronic device includes a processor and a memory storing processor-executable instructions, which, when executed by the processor, implement the method as described in any one of claims 1 to 10.

13. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by a processor, it implements the method as described in any one of claims 1 to 10.

Images