Early warning method, device and equipment based on driver attention and storage medium

Abstract

The embodiment of the invention provides an early warning method and device based on driver attention, equipment and a storage medium, and the method comprises the steps: obtaining the sight line information of a driver in a preset time window, and determining the attention information corresponding to the preset time window according to the sight line information in the preset time window; wherein the attention information comprises an attention area and area brightness, the attention area represents an area concerned by the driver, and the area brightness represents the attention degree of the driver to the attention area; determining the attention information corresponding to the preset time period according to the attention information corresponding to the plurality of preset time windows in the preset time period; if it is recognized that the object exists in the attention area corresponding to the preset time period, determining a strategy for sending early warning information to the driver according to the type information of the object and the area brightness corresponding to the preset time period; wherein the early warning information is used for reminding a driver to pay attention to the object.

Description

Technical Field

[0001] This application relates to the field of intelligent vehicles, and more particularly to a warning method, device, equipment, and storage medium based on driver attention. Background Technology

[0002] With the development of intelligent driving technology, the driver's line of sight can be collected in real time through the DMS (Driver Monitoring System). Based on the driver's line of sight, the system can identify whether there are obstacles outside the vehicle, thereby triggering a warning.

[0003] However, such methods are prone to false alarms or missed alarms for obstacles, and cannot accurately warn of objects based on the driver's attention, affecting the efficiency of human-computer interaction and driving safety. Summary of the Invention

[0004] This application provides a driver attention-based warning method, device, equipment, and storage medium to provide targeted warnings of objects based on the driver's attention, thereby improving driving safety.

[0005] In a first aspect, embodiments of this application provide a warning method based on driver attention, comprising:

[0006] The driver's gaze information within a preset time window is acquired, and attention information corresponding to the preset time window is determined based on the gaze information within the preset time window; wherein, the attention information includes an attention region and a region brightness, the attention region representing the area that the driver is paying attention to, and the region brightness representing the degree of attention the driver pays to the attention region;

[0007] Based on the attention information corresponding to multiple preset time windows within a preset time period, determine the attention information corresponding to the preset time period; and

[0008] If an object is detected in the attention area corresponding to the preset time period, a strategy for sending a warning message to the driver is determined based on the object's type information and the brightness of the area corresponding to the preset time period; wherein, the warning message is used to remind the driver to pay attention to the object.

[0009] Secondly, embodiments of this application provide a warning device based on driver attention, comprising:

[0010] The first determining unit is used to acquire the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window; wherein, the attention information includes an attention region and a region brightness, the attention region representing the area that the driver is paying attention to, and the region brightness representing the degree of attention the driver pays to the attention region;

[0011] The second determining unit is configured to determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within a preset time period; and

[0012] An object detection unit is configured to determine a strategy for sending a warning message to the driver based on the object's type information and the brightness of the area corresponding to the preset time period if an object is detected in the attention area corresponding to the preset time period; wherein the warning message is used to remind the driver to pay attention to the object.

[0013] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0014] The memory stores computer-executed instructions;

[0015] The processor executes computer execution instructions stored in the memory, causing the processor to perform the implementation method described in the first aspect above.

[0016] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the embodiments described in the first aspect above.

[0017] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the implementation methods described in the first aspect above.

[0018] This application provides a driver attention-based warning method, device, equipment, and storage medium. It acquires the driver's gaze information within a preset time window, and determines the attention region and its brightness based on this gaze information. In other words, it determines the area the driver focuses on and the degree of attention they pay to that area within the preset time window, achieving dynamic recognition of driver attention. It fuses attention information from multiple preset time windows to obtain attention information corresponding to a preset time period. That is, it uses attention regions and their brightness from multiple short time windows to form a long-period attention region and its brightness, improving the temporal continuity and stability of attention judgment. It performs object recognition on the attention region corresponding to the preset time period to determine the object's type information. Based on the object's type information and the brightness of the region corresponding to the preset time period, it determines a strategy for sending warning information to the driver, reminding the driver to pay targeted attention to the object, improving the accuracy of object warnings, realizing a differentiated prompting mechanism, and enhancing the safety and comfort of human-computer interaction. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] Figure 1 A flowchart illustrating a driver attention-based early warning method provided in this application embodiment;

[0021] Figure 2 A schematic diagram of line-of-sight information provided in an embodiment of this application;

[0022] Figure 3 A flowchart illustrating a driver attention-based early warning method provided in this application embodiment;

[0023] Figure 4 A schematic diagram of the attention region provided in the embodiments of this application;

[0024] Figure 5 A flowchart illustrating a driver attention-based early warning method provided in this application embodiment;

[0025] Figure 6 A schematic diagram of the attention region corresponding to the preset time period provided in the embodiments of this application;

[0026] Figure 7 This is a schematic diagram of the brightness attenuation relationship information provided in the embodiments of this application;

[0027] Figure 8 A flowchart illustrating a driver attention-based early warning method provided in this application embodiment;

[0028] Figure 9 A schematic diagram of a driver attention-based warning device provided in an embodiment of this application;

[0029] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0030] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0032] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0033] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0034] In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0035] It should be noted that, due to space limitations, this application specification does not exhaustively list all possible implementation methods. Those skilled in the art, after reading this application specification, should be able to deduce that, as long as the technical features do not contradict each other, any combination of technical features can constitute an optional implementation method. The following provides a detailed description of each embodiment.

[0036] Currently, in practical applications, driver assistance systems (ADAS) typically rely on sensors to detect objects in the road environment and issue warnings to the driver based on preset rules. However, these systems often lack the ability to perceive the driver's own attention state, making it difficult to determine whether the driver has noticed potential hazards ahead. For example, during high-speed driving, even if ADAS detects a pedestrian or obstacle ahead, if the driver's eyes are already focused on that area, repeated warnings not only fail to improve safety but may also interfere with driving attention and cause cognitive overload. Conversely, if the driver is not paying attention to high-risk targets, the system may fail to provide timely warnings, potentially leading to delayed reactions. Therefore, accurately identifying the driver's actual area of ​​focus and the intensity of that focus has become a key issue in optimizing human-machine interaction and improving the effectiveness of warnings.

[0037] This application provides a driver attention-based early warning method, device, equipment, and storage medium, which aims to solve the above-mentioned technical problems of the prior art.

[0038] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0039] Figure 1 This is a flowchart illustrating a driver attention-based warning method provided in an embodiment of this application. This method can be executed by a driver attention-based warning device. Figure 1 As shown, the method includes:

[0040] S101. Obtain the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window; wherein, the attention information includes the attention area and the area brightness, the attention area represents the area that the driver is paying attention to, and the area brightness represents the degree of attention the driver pays to the attention area.

[0041] For example, a preset time window refers to a time segment used to collect and analyze the driver's instantaneous visual behavior. Its length can be flexibly set according to the application scenario; for example, the preset time window can be 1 second. Shorter time windows are suitable for capturing rapid eye movement changes, while longer windows help smooth out noise and reflect sustained gaze trends. This time window can be fixed or dynamically adjusted according to vehicle operating conditions such as vehicle speed and curve curvature.

[0042] The driver's gaze information within a preset time window can be collected via a visual monitoring system (DMS). For example, if the DMS operates at a frequency of 5Hz, it can collect five gaze information points per second. Gaze information refers to data on the driver's eye movements collected by the DMS, specifically including the gaze direction vector, pupil position, head posture, and other information at each moment. For instance, the gaze direction vector can be extracted using an infrared camera combined with image processing algorithms. In this embodiment, the method for collecting gaze information is not specifically limited. Figure 2 This is a diagram illustrating line-of-sight information. Figure 2 In the diagram, a circular pattern represents the driver, who collects five line-of-sight information points within a preset time window, which then radiate outwards in the form of rays.

[0043] For each preset time window, the driver's attention information corresponding to that time window can be determined based on all line-of-sight information within that window. Attention information can include the attention region and the brightness of that region. The attention region represents the area of ​​vision the driver is focused on, i.e., the spatial range of attention derived from line-of-sight information. The attention region can be represented in various geometric forms, such as rectangles, polygons, sectors, or Gaussian heatmaps. In this embodiment, the attention region is modeled as a sector centered on the main line-of-sight direction, and the angle and radius of this sector can be determined by the dispersion of the line of sight and depth estimation.

[0044] Zone brightness is a quantitative indicator of attention concentration, reflecting the intensity of a driver's focus on content within a specific area of ​​attention. The numerical range of zone brightness can be set to [0, 1], with higher values ​​indicating greater concentration. Zone brightness can be calculated by comprehensively considering gaze duration, gaze fluctuation amplitude, blink frequency, and other gaze-related information. For example, in one embodiment, zone brightness is negatively correlated with gaze variance; more stable gaze, i.e., smaller variance, results in higher zone brightness; conversely, frequent saccades increase variance and decrease zone brightness.

[0045] The process of determining attention information can involve extracting features from raw gaze information and mapping them to attention information. For example, the received continuous gaze information can be filtered and denoised, and then the mean and variance can be calculated within a preset time window. The extracted features, combined with a preset mapping function or machine learning model, can be converted into specific attention information, thus realizing a calculable description of the driver's subjective attention state.

[0046] This embodiment transforms raw gaze signals into structured attention states. By introducing a dual representation mechanism of attention area and area brightness, it can not only identify where the driver is looking, but also determine how focused the driver is, thus more accurately distinguishing between brief glances and genuine attention. This lays the foundation for subsequent cross-timescale information fusion and contextualized prompts, improving the intelligence level of warnings and user experience.

[0047] S102. Determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within the preset time period.

[0048] For example, a preset time period is a time interval longer than a single time window, used to integrate short-term attentional states and generate periodic attentional information. A preset time period may include multiple preset time windows, and the last moment in the preset time period may be the current moment.

[0049] Multiple preset time windows are distributed within a preset time period, and can be arranged consecutively, overlappingly, or non-overlappingly. For example, if each second is divided into a time window, then a 10-second period contains 10 preset time windows. Each preset time window independently generates its own attention information, which is then uniformly incorporated into the aggregation process.

[0050] The aggregation process aims to integrate multiple attention information sets to generate a more robust and representative overall attention description. Specifically, it identifies all regions of interest to the user within a preset time period, defining them as the attention regions corresponding to that time period. This can be achieved, for example, by calculating the union, intersection, or cluster centers of the attention regions for each preset time window. Furthermore, it calculates the overall region brightness, using methods such as weighted averaging or maximum value retention. In other words, it obtains the attention regions and region brightness corresponding to the preset time period, serving as the attention information for that period.

[0051] This embodiment upgrades attention from instantaneous to periodic attention. By employing a multi-time window fusion mechanism, it effectively reduces the risk of misjudgment caused by single sampling errors, improving the stability and reliability of attention state recognition. This is particularly important for safety warnings in complex traffic scenarios, ensuring that alert decisions are based on sufficient and continuous behavior.

[0052] S103. If an object is detected in the attention area corresponding to the preset time period, a strategy for sending a warning message to the driver is determined based on the object type information and the brightness of the area corresponding to the preset time period; wherein, the warning message is used to remind the driver to pay attention to the object.

[0053] For example, objects refer to road users or static obstacles detected by the vehicle-mounted perception system, including but not limited to motor vehicles, non-motor vehicles, pedestrians, animals, construction signs, traffic lights, etc. Information such as the position, type, and speed of these objects is typically provided by lidar, millimeter-wave radar, cameras, and multi-sensor fusion algorithms. In this embodiment, the object recognition algorithm is not specifically limited. Object type information is an identifier used to distinguish different object categories, which can be represented by semantic tags, priority levels, or standardized codes, etc.

[0054] After determining the attention area corresponding to a preset time period, the system can detect the presence of objects within that area in real time. If no object is found, the detection continues; if an object is found, its type information is determined. Based on the object's type information and the brightness of the area corresponding to the preset time period, a strategy for sending warning information to the driver can be determined. For example, it can be determined whether a warning information needs to be issued, as well as the content and delivery method of the warning information.

[0055] The brightness of the area corresponding to the preset time period can serve as a reference benchmark for the driver's attention level, used to assess the driver's potential probability of perceiving objects within that area. Thresholds for each type of information are preset, and these thresholds are compared with the area brightness. If the threshold for an object is less than the area brightness, it indicates that the driver has likely already noticed it, and no warning is needed, or the warning intensity can be reduced; conversely, if the threshold is greater than the area brightness, it indicates that the object poses a risk of being ignored by the driver, requiring an emergency warning.

[0056] Warning information is the carrier of information that warns drivers, and its presentation can include audio prompts, visual cues, tactile feedback, etc. The prompting strategy can adaptively adjust according to the situation; for example, strong, multimodal prompts are used in emergencies, while gentle reminders are used in normal situations.

[0057] The core logic for determining whether to send a warning message lies in comparing the "degree of attention an object should receive" with the "actual level of driver attention." This achieves an intelligent prompting mechanism based on attention matching, enabling differentiated responses to different situations. It avoids redundant alarms for already noticed targets and prevents the omission of high-risk objects. This significantly improves human-machine collaboration efficiency, reduces unnecessary interference, and enhances the usability and reliability of the driver assistance system.

[0058] This application provides a driver attention-based warning method. It acquires the driver's gaze information within a preset time window, and determines the corresponding attention region and its brightness based on this gaze information. In other words, it determines the area the driver focuses on within the preset time window and the degree of attention paid to that area, achieving dynamic recognition of driver attention. Attention information from multiple preset time windows is fused to obtain attention information corresponding to a preset time period. That is, by using attention regions and their brightness from multiple short time windows, a long-period attention region and its brightness are formed, improving the temporal continuity and stability of attention judgment. Object recognition is performed on the attention region corresponding to the preset time period to determine the object's type information. Based on the object's type information and the corresponding brightness of the region within the preset time period, a strategy for sending warning information to the driver is determined, reminding the driver to pay targeted attention to the object, improving the accuracy of object warnings, realizing a differentiated prompting mechanism, and enhancing the safety and comfort of human-computer interaction.

[0059] Figure 3 A flowchart illustrating a driver attention-based warning method provided in this application embodiment is shown below. Figure 3 As shown, in this embodiment... Figure 1 Based on the embodiments, a warning method based on driver attention is described in detail, the method comprising:

[0060] S301. Obtain the driver's gaze information within a preset time window, and determine the gaze characteristics corresponding to the preset time window based on the gaze information within the preset time window; wherein, the gaze characteristics represent the driver's main gaze direction and the degree of dispersion of gaze within the preset time window.

[0061] For example, driver gaze information can be acquired periodically according to a preset time window. The gaze information within the preset time window refers to the driver's gaze direction data at different times during a continuous period of time, collected by an onboard eye tracker or other vision tracking device, reflecting the spatial orientation of the driver's instantaneous gaze.

[0062] Visual gaze characteristics are parameters extracted from raw visual gaze information through statistical analysis. They refer to transforming discrete, fluctuating sequences of visual gaze directions into quantitative indicators that characterize overall trends and distributional properties. Visual gaze characteristics include at least two dimensions: the primary visual gaze direction and the degree of visual gaze dispersion. The former reflects the driver's average focus direction within a preset time window, while the latter reflects whether the driver's attention is concentrated or scattered.

[0063] In this embodiment, the main line-of-sight direction can be obtained by calculating the mean of all line-of-sight directions, and a weighted average method in a spherical coordinate system is used to handle periodic deviations; the degree of line-of-sight dispersion can be measured by calculating the variance, standard deviation, etc. between directions. In this embodiment, no specific limitation is made on the method for extracting line-of-sight features.

[0064] This embodiment extracts gaze feature parameters that represent the overall trend by statistically analyzing gaze information within a preset time window. This solves the problem that the original gaze information has high data noise and is difficult to directly reflect the driver's attention distribution trend, thereby improving the robustness and interpretability of attention recognition.

[0065] In this embodiment, the gaze information within a preset time window includes the direction of gaze at multiple moments within the preset time window. Based on the gaze information within the preset time window, determining the gaze characteristics corresponding to the preset time window includes: performing mean processing on the gaze information within the preset time window to obtain a first feature corresponding to the preset time window; wherein the first feature characterizes the driver's primary gaze direction within the preset time window; performing variance processing on the gaze information within the preset time window to obtain a second feature corresponding to the preset time window; wherein the second feature characterizes the degree of dispersion of the driver's gaze within the preset time window; and determining the gaze characteristics based on the first and second features.

[0066] Specifically, the preset time window refers to a time segment used to collect and analyze the driver's gaze behavior. The preset time window includes multiple moments, each corresponding to the collected gaze information, which indicates the driver's gaze direction at that moment. In other words, the gaze information within the preset time window is continuously collected by the onboard eye tracker or visual perception module, and the gaze information at each moment represents the spatial direction of the driver's gaze at that moment.

[0067] After obtaining the gaze information at various moments within a preset time window, this gaze information is averaged. Averaged processing involves performing an arithmetic mean operation on the gaze direction data at all sampling moments within the preset time window to obtain a direction parameter representing the overall trend, i.e., the first feature, denoted as μ. The first feature reflects the driver's average gaze direction within the preset time window. For example, in a straight-ahead driving scenario, if most gaze directions are concentrated in the center of the road ahead, the average result will tend towards the direction directly ahead; while during lane changes, the average may be biased towards the side mirrors or adjacent lane areas.

[0068] After obtaining the gaze information at various moments within a preset time window, variance processing can be applied to this gaze information. Variance processing quantifies the fluctuation of the gaze direction within the preset time window, i.e., the degree of deviation of each sampling point from the mean direction. This statistic constitutes the second feature, denoted as σ, reflecting the concentration or dispersion of the driver's visual attention. A smaller variance indicates a more stable gaze and concentrated attention; a larger variance indicates frequent scanning or switching between multiple targets, and a tendency for attention to be scattered. In this embodiment, other dispersion measures can also be used instead of variance, such as mean absolute deviation, interquartile range, or entropy, which are particularly suitable for modeling non-Gaussian distributed gaze data.

[0069] The gaze feature is composed of a first feature and a second feature, used to characterize the driver's visual behavior pattern within a preset time window. The gaze feature not only includes the direction of attention, i.e., the main gaze direction, but also incorporates the stability of attention, i.e., the degree of gaze dispersion, and can be characterized as (μ, σ).

[0070] The beneficial effect of this setting is that it enables effective modeling of driver gaze behavior. By processing the gaze information within a preset time window using mean and variance, the first feature representing the main gaze direction and the second feature representing the degree of gaze dispersion are extracted respectively. Based on the joint determination of the two features, the gaze characteristics are provided, which provides a reliable data foundation for the subsequent construction of accurate attention areas and brightness distribution, thereby supporting more intelligent driving assistance decision-making and warning mechanisms.

[0071] S302. Determine the attention information corresponding to the preset time window based on the gaze characteristics corresponding to the preset time window.

[0072] For example, the gaze characteristics include two core parameters: the main gaze direction and the degree of gaze dispersion, which correspond to the spatial orientation and focus of attention, respectively.

[0073] Attention information is a higher-level cognitive state expression derived from gaze characteristics, specifically comprising two components: the attention region and the region's brightness. The attention region is a spatial geometric range, such as a sector, ellipse, or polygon, used to define the area of ​​the driver's forward field of vision that is most likely to be focused on; the region's brightness is a scalar value or spatial function used to quantify the intensity of the driver's attention to that area.

[0074] In this embodiment, the shape of the attention area can be dynamically adjusted according to the characteristics of the gaze: when the main gaze direction is stable and the dispersion is low, a small and clearly oriented fan-shaped area is generated; when the dispersion is high, the opening angle is increased or a more diffuse area shape, such as a circle, is adopted. The brightness of the area can be directly mapped inversely to the dispersion of the gaze. The smaller the dispersion, the more concentrated the attention, and the higher the brightness of the corresponding area; conversely, the higher the dispersion, the lower the brightness.

[0075] This embodiment generates structured attention information based on gaze characteristics, which solves the problem that directly modeling attention from raw gaze information is easily affected by jitter and cannot accurately depict the range and intensity of attention, thereby enhancing the accuracy of object warning.

[0076] In this embodiment, the gaze characteristics include a first feature and a second feature. The first feature represents the driver's main gaze direction within a preset time window, and the second feature represents the degree of dispersion of the driver's gaze within the preset time window. Based on the gaze characteristics corresponding to the preset time window, the attention information corresponding to the preset time window is determined, including: determining the attention region corresponding to the preset time window based on the first and second features corresponding to the preset time window; and determining the region brightness corresponding to the preset time window based on the second feature corresponding to the preset time window.

[0077] Specifically, the first feature represents the driver's main gaze direction within a preset time window. This main gaze direction is a direction vector obtained by averaging the driver's gaze direction at multiple consecutive moments, which is used to reflect the driver's average gaze direction within the preset time window.

[0078] The second characteristic characterizes the dispersion of the driver's gaze within a preset time window, typically quantified using statistical measures such as variance, standard deviation, or covariance matrix. When the driver maintains a stable gaze on a fixed target, the gaze direction is highly concentrated at all times, resulting in a smaller second characteristic value; conversely, when the driver frequently scans multiple areas or their attention is scattered, the gaze direction changes drastically, leading to a larger second characteristic value. This dispersion not only reflects the spatial distribution of visual attention but also indirectly reflects the level of cognitive focus.

[0079] Based on the first and second features corresponding to the preset time window, the attention region corresponding to the preset time window can be determined. The attention region refers to the spatial range in the driver's forward field of vision that is determined to be the current focus, and its shape can be modeled as a fan, ellipse, or polygon, etc. In this embodiment, a fan shape is used to model the attention region, with the center located at the driver's eye position, the opening direction determined by the first feature, and the main line of sight as the direction of the central axis of the fan. The size of the fan's angle is controlled by the second feature. When the second feature value is low, that is, when the line of sight is highly concentrated, the angle is set to a smaller range, indicating that the attention is focused on a narrow area; when the second feature value is high, that is, when the line of sight is dispersed, the angle is expanded, indicating that the attention covers a wider field of vision. In addition, the radius of the fan can be dynamically adjusted according to other factors such as vehicle speed and environment, or it can be set to a fixed distance to simplify implementation.

[0080] Regional brightness is used to characterize the strength of a driver's attention to a particular area; it is a dimensionless numerical value or a normalized intensity index. In one embodiment, regional brightness and the second feature can be negatively correlated, meaning that the more focused the gaze, the smaller the second feature value and the higher the regional brightness; conversely, the more dispersed the gaze, the lower the regional brightness. This can be achieved by establishing a mapping function, which calculates the corresponding regional brightness based on a preset mapping function.

[0081] The first feature provides spatial orientation information of attention, while the second feature simultaneously affects the spatial expansion of the attention region and the intensity of the region's brightness. Together, they constitute a two-dimensional description of the driver's attention state, achieving a reasonable mapping from raw eye-tracking data to psychological perception.

[0082] The beneficial effect of this setup is that it transforms abstract visual characteristics into a concrete representation of attention space, more realistically restoring the driver's attention during driving and improving the accuracy of subsequent object warnings.

[0083] In this embodiment, determining the attention region corresponding to the preset time window based on the first and second features corresponding to the preset time window includes: determining region direction information based on the first feature corresponding to the preset time window; wherein the region direction information represents the direction of the fan-shaped opening of the attention region; determining central angle information based on the second feature corresponding to the preset time window; wherein the central angle information represents the range of the fan-shaped opening of the attention region; determining radius information based on the line-of-sight information within the preset time window; wherein the radius information represents the radius of the fan-shaped opening of the attention region; and determining the attention region corresponding to the preset time window based on the region direction information, central angle information, and radius information.

[0084] Specifically, the first feature is obtained by averaging the gaze directions at multiple moments within a preset time window, used to characterize the driver's primary gaze direction during that time period. This primary gaze direction can be mapped to an angle value in a two-dimensional coordinate system or a polar coordinate system, such as an angle representation with 0° as the vehicle's forward direction as the reference, left being positive and right being negative, or expressed as a unit vector in a Cartesian coordinate system. The regional direction information is determined based on this primary gaze direction, serving as the direction of the central axis of the fan-shaped structure formed by the attention area. That is, the regional direction information characterizes the direction of the fan-shaped opening of the attention area. For example, if the driver's primary gaze direction within the preset time window is concentrated 15° to the left of the front, the regional direction information is set to 15°, indicating that the attention area expands around this angle. The regional direction information is not limited to a single instantaneous direction, but reflects a stable gaze trend over a period of time, thereby avoiding misjudgments caused by brief eye movements.

[0085] The second feature, obtained by variance processing of the gaze direction within a preset time window, quantifies the dispersion of the gaze, i.e., the level of concentration or dispersion of the driver's gaze distribution within the preset time window. The central angle information, also known as the fan-shaped angle, is mapped from the second feature and defines the lateral opening angle of the attention area. For example, in low variance cases, the system can set the central angle to a smaller value, indicating that attention is concentrated within a narrow field of vision; while in high variance cases, the central angle can expand to 60° or even larger, reflecting that the driver is conducting a broad environmental scan.

[0086] Radius information reflects the distance the driver's attention extends in the depth dimension, i.e., the spatial depth of their area of ​​focus. Radius information can be determined from the depth perception component of gaze information, specifically by: calculating the distance to the gaze focus using the eye convergence angle collected by binocular cameras; or predicting the gaze depth by combining monocular vision with machine learning models; or by using external sensors such as LiDAR or millimeter-wave radar to assist in retrieving the location of the gaze point.

[0087] In one implementation, the system analyzes the position of the driver's pupils in real time and calculates the distance in front of the intersection of the lines of sight between the two eyes, which is used as the depth of vision value. The system then averages or weights the depth of vision values ​​at all times within a preset time window to obtain a representative depth estimate, which serves as the radius information of the sector. The radius information can also be set to a fixed empirical value, such as 50 meters, or dynamically adjusted according to the vehicle's current speed; the higher the speed, the larger the radius, reflecting the driver's need to pay attention to distant targets in advance.

[0088] The attention region is ultimately constructed into a two-dimensional sector-shaped spatial region by the three parameters mentioned above, which is the attention region corresponding to the preset time window. The attention is defined with the driver's eye as the vertex, the region direction information as the central axis, the central angle information as the included angle width, and the radius information as the maximum extension length, which fully describes the driver's visual attention range within the preset time window. Figure 4 This is a schematic diagram of the attention area. Figure 4 In the diagram, θ represents the central angle, R represents the radius, and the arrow indicates the direction of the region.

[0089] The advantage of this setup is that it transforms abstract attention into quantifiable geometric regions, enabling a more realistic simulation of human visual attention distribution and improving the rationality and personalization of object-based cueing decisions.

[0090] In this embodiment, determining the radius information based on the line-of-sight information within a preset time window includes: determining the line-of-sight depth value based on the line-of-sight information within the preset time window; wherein the line-of-sight depth value represents the extension distance of the driver's line of sight; and determining the radius information based on the line-of-sight depth value and a preset depth threshold.

[0091] Specifically, the preset time window includes multiple gaze information points. By analyzing the driver's gaze information at multiple consecutive moments, such as binocular vision data or eye focusing state, the actual projection depth of the driver's gaze in three-dimensional space is estimated. The gaze depth value reflects the spatial distance of the driver's current focus point and is one of the key parameters for constructing the three-dimensional range of the attention area. For example, when the driver is looking at a traffic light several hundred meters ahead while the vehicle is in motion, their eyes converge at a small angle but are clearly focused. The system can collect changes in interpupillary distance and lens accommodation signals through an onboard eye tracker and deduce the gaze depth value by combining it with a calibrated visual model. As an optional implementation, a deep learning-based monocular depth estimation algorithm can also be used to predict the approximate distance of the driver's gaze target using a sequence of facial images, thereby obtaining an approximate estimate of the gaze depth value.

[0092] Based on the line-of-sight depth value and a preset depth threshold, radius information is determined. This radius represents the final numerical value used to define the fan-shaped radius of the attention region. The radius information is not directly derived from the original line-of-sight depth value, but rather is the result of a comprehensive decision after comparing the line-of-sight depth value with a preset depth threshold. For example, the smaller of the two values ​​can be used as the final radius information to prevent misjudgment from causing the attention region to be over-extrapolated to unreliable distances. For instance, if the calculated line-of-sight depth value is 150m and the preset depth threshold is 80m, the final determined radius information will be 80m, thus limiting the attention region from exceeding the safe perception range.

[0093] The beneficial effect of this setting is that the line-of-sight depth value provides individualized and real-time attention depth estimation, while the preset depth threshold introduces a safety constraint mechanism. The two work together to generate radius information, so that the attention area has both personalized perception capabilities and conforms to the overall driving safety logic, thereby improving the response accuracy and safety of the driving assistance system.

[0094] In this embodiment, the method further includes: acquiring current driving information; wherein the driving information includes at least vehicle speed information and environmental information; determining a threshold corresponding to the current driving information based on a preset first correlation relationship, which is a preset depth threshold; wherein the preset first correlation relationship characterizes the correlation relationship between the driving information and the threshold.

[0095] Specifically, the preset depth threshold can dynamically change according to the vehicle's driving conditions and the external environment. Current driving information is acquired, including at least vehicle speed and environmental information. For example, parameters related to the vehicle's operating status and the surrounding driving environment can be collected in real time through onboard sensors, communication modules, or external data sources. Vehicle speed is a crucial factor influencing the driver's attention distribution: at high speeds, drivers focus more on distant areas to cope with unexpected road conditions; at low speeds, they tend to focus on nearby details, such as pedestrians and obstacles. Therefore, vehicle speed information directly affects the reasonable range of the depth of vision value. Environmental information encompasses various external condition parameters, including but not limited to weather conditions, light intensity, road type, traffic density, and visibility level. Environmental information can originate from onboard cameras, millimeter-wave radar, weather sensors, or cloud-based map services. For example, in rainy or foggy weather, moisture in the air reduces the visibility of distant objects, making it difficult for the driver to focus on distant areas. In such cases, the depth threshold of the attention area should be appropriately reduced to avoid misjudging blurred distant scenery as high-attention targets.

[0096] The preset first association is a pre-established mapping logic or function model used to describe the quantitative or qualitative correspondence between driving information and depth thresholds. This association can be constructed through experimental calibration, machine learning training, or expert experience rules. The depth threshold is used to define the maximum extension distance of the attention area. Based on the preset first association, the depth threshold corresponding to the current driving information is found. If the line-of-sight depth value exceeds the depth threshold, the depth threshold is determined as radius information; otherwise, the line-of-sight depth value is radius information.

[0097] For example, when the detected vehicle speed is greater than 80 km / h and the environment is a sunny highway, the system calls the depth threshold of 150 meters; when the vehicle speed is less than 30 km / h and the environment is a rainy city, the system calls the depth threshold of 60 meters.

[0098] The beneficial effects of this setting are that it enables dynamic adjustment of the depth threshold of the attention area based on the actual driving state, taking into account diverse driving scenarios, improving the accuracy and reliability of object recognition prompts, and enhancing the safety and user experience of human-computer interaction.

[0099] S303. Determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within the preset time period.

[0100] For example, this step can refer to step S102 above, and will not be repeated here.

[0101] S304. If an object is detected in the attention area corresponding to a preset time period, a strategy for sending a warning message to the driver is determined based on the object type information and the brightness of the area corresponding to the preset time period; wherein, the warning message is used to remind the driver to pay attention to the object.

[0102] For example, this step can refer to step S103 above, and will not be repeated here.

[0103] This application provides a driver attention-based warning method. It acquires the driver's gaze information within a preset time window, and determines the corresponding attention region and its brightness based on this gaze information. In other words, it determines the area the driver focuses on within the preset time window and the degree of attention paid to that area, achieving dynamic recognition of driver attention. Attention information from multiple preset time windows is fused to obtain attention information corresponding to a preset time period. That is, by using attention regions and their brightness from multiple short time windows, a long-period attention region and its brightness are formed, improving the temporal continuity and stability of attention judgment. Object recognition is performed on the attention region corresponding to the preset time period to determine the object's type information. Based on the object's type information and the corresponding brightness of the region within the preset time period, a strategy for sending warning information to the driver is determined, reminding the driver to pay targeted attention to the object, improving the accuracy of object warnings, realizing a differentiated prompting mechanism, and enhancing the safety and comfort of human-computer interaction.

[0104] Figure 5 A flowchart illustrating a driver attention-based warning method provided in this application embodiment is shown below. Figure 5 As shown, this embodiment, based on the above embodiments, provides a detailed description of a warning method based on driver attention. The method includes:

[0105] S501. Obtain the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window; wherein, the attention information includes the attention area and the area brightness, the attention area represents the area that the driver is paying attention to, and the area brightness represents the degree of attention the driver pays to the attention area.

[0106] For example, this step can refer to step S101 above, and will not be repeated here.

[0107] S502. Determine the attention area corresponding to the preset time period based on the attention area corresponding to each preset time window.

[0108] For example, the preset time period includes multiple preset time windows, each corresponding to its own attention region. The attention region corresponding to the preset time period is determined by combining the attention regions corresponding to each preset time window. That is, based on the spatial distribution characteristics of the driver's attention across multiple consecutive or non-consecutive time windows, spatial aggregation analysis is performed to form an overall attention region that reflects the driver's sustained focus over a longer time scale. This overall attention region includes not only overlapping areas repeatedly glanced at by the driver but also non-overlapping areas briefly scanned, thus providing a more comprehensive characterization of the driver's visual state. For instance, during highway driving, the driver may frequently scan the center of the lane ahead and the edges of adjacent lanes within multiple time windows; spatial aggregation can identify that their primary focus is on the driving path ahead and its surrounding area.

[0109] Figure 6 This is a schematic diagram of the attention area corresponding to a preset time period. Figure 6 In the preset time period, there are two preset time windows. The attention regions of the two preset time windows are combined, and the union of the combined regions is determined as the attention region of the preset time period.

[0110] In this embodiment, determining the attention region corresponding to a preset time period based on the attention region corresponding to each preset time window includes: determining at least one sub-region based on the attention region corresponding to each preset time window; wherein, the sub-region includes overlapping regions and / or non-overlapping regions, the overlapping region represents the intersection of the attention regions between at least two preset time windows, and the non-overlapping region represents the region other than the overlapping region in the attention region corresponding to the preset time window; and determining at least one sub-region as the attention region corresponding to the preset time period.

[0111] Specifically, the attention region corresponding to the preset time window can be a sector in a two-dimensional image coordinate system, reflecting the driver's main visual focus range within a specific time period. By performing spatial set operations on the attention regions of multiple preset time windows, one or more sub-regions can be determined. For example, by taking different attention regions as geometric object inputs and using set operations such as intersection, union, and difference, sub-regions can be divided.

[0112] A sub-region can be an overlapping area between different attention regions or a non-overlapping area. An overlapping area refers to the spatial intersection of the attention regions of two or more preset time windows. For example, in three consecutive preset time windows T1, T2, and T3, if the attention regions are A1, A2, and A3 respectively, and A1∩A2≠ A2∩A3= The intersection of A1 and A2 constitutes the overlapping region, and the portion outside the overlapping region is the non-overlapping region. Each preset time window may correspond to a non-overlapping region, such as... Figure 6 As shown, Figure 6 This includes one overlapping region and two non-overlapping regions. The overlapping region typically reflects key areas that the driver repeatedly glances at or continuously focuses on, such as vehicles ahead, traffic lights, or blind spots during lane changes. The non-overlapping region refers to the portion of the attention area within a pre-defined time window that is not covered by other pre-defined time windows; that is, it belongs to the attention range of a single fixation behavior. This type of region can be used to identify momentary distractions or sudden shifts in attention.

[0113] After obtaining multiple sub-regions, all sub-regions are identified as attention regions corresponding to preset time periods. That is, instead of simply merging or averaging attention regions across multiple preset time windows, the differences in their internal structures are preserved, forming a composite attention distribution map composed of multiple sub-regions. This representation method can simultaneously reflect the spatial distribution characteristics and temporal evolution patterns of attention.

[0114] Each sub-region, as an independent information unit, can be assigned different brightness levels in subsequent processing and participate in the object detection and prompting decision-making process. For example, if an obstacle is located within an overlapping area, it indicates that the driver has noticed the target multiple times, and the system can accordingly lower the prompt level; conversely, if the obstacle appears in a non-overlapping area or is not covered by any sub-region, the prompt level can be increased accordingly.

[0115] The beneficial effect of this setting is that it enables a refined reconstruction of the attention regions of multiple preset time windows. By identifying overlapping and non-overlapping regions, a sub-region system containing spatiotemporal evolution information is constructed, which solves the problems of information redundancy and loss of details in the process of multi-time period attention fusion, and provides a reliable basis for subsequent differentiated prompting strategies.

[0116] S503. For each preset time window within a preset time period, adjust the brightness of the area corresponding to the preset time window to obtain the current brightness corresponding to the preset time window.

[0117] For example, for each preset time window within a preset time period, the preset time window becomes the historical time whenever the regional brightness of that preset time window is determined. The regional brightness of the attention area in a single time window gradually decreases over time; therefore, a time decay mechanism is introduced to dynamically adjust the historical regional brightness.

[0118] Regional brightness, as a quantitative indicator of a driver's attention to a specific spatial area, is initially derived from characteristics such as gaze stability and dwell time within a single time window. However, as time progresses, the contribution of early-determined regional brightness to current decision-making gradually decreases. Therefore, regional brightness can be scaled in real-time or periodically to obtain current brightness that reflects timeliness. For example, the further away a preset time window is from the current moment, the smaller its corresponding regional brightness, indicating that the attention behavior within that preset time window has a weaker impact on the current attention state.

[0119] In this embodiment, adjusting the brightness of the area corresponding to the preset time window to obtain the current brightness corresponding to the preset time window includes: determining the brightness attenuation coefficient of the preset time window according to preset brightness attenuation relationship information; wherein, the brightness attenuation relationship information characterizes the relationship between the brightness attenuation coefficient and time, and the brightness attenuation coefficient characterizes the degree of brightness attenuation of the area; adjusting the brightness of the area corresponding to the preset time window according to the brightness attenuation coefficient of the preset time window to obtain the current brightness corresponding to the preset time window.

[0120] Specifically, brightness attenuation relationship information refers to a pre-defined function model or lookup table used to describe the mapping relationship between time and the brightness attenuation coefficient. This relationship can be linear attenuation, exponential attenuation, or other non-linear forms. The brightness attenuation coefficient characterizes the degree of brightness attenuation in a region. Figure 7 This is a schematic diagram illustrating the relationship between brightness decay. Figure 7 In the diagram, the horizontal axis represents time t, and the vertical axis represents the brightness attenuation coefficient f. [0, 1], the closer to the current time, the closer the brightness attenuation coefficient is to 0, that is, the lower the brightness of the area.

[0121] Adjusting the brightness of a region means multiplying the original detected brightness of the region by the corresponding brightness attenuation coefficient to obtain a time-weighted current brightness value.

[0122] The beneficial effect of this setting is that by introducing a brightness decay mechanism based on time variables, the influence of longer-term attention behaviors on current decisions gradually weakens, thereby improving the timeliness and accuracy of attention state assessment, enhancing the intelligence level of object cues and the efficiency of human-machine collaboration.

[0123] S504. Determine the area brightness corresponding to the preset time period based on the current brightness corresponding to each preset time window.

[0124] For example, all the attenuated current brightness values ​​are integrated to obtain the area brightness corresponding to a preset time period. The integration method can be the maximum value, weighted average, or integral accumulation. For example, in application scenarios that need to highlight the strongest focus, the highest current brightness can be selected as the final area brightness; while when evaluating overall attention, a weighted summation method is more suitable to comprehensively reflect the average attention level over the entire period.

[0125] This embodiment achieves temporal-spatial dual fusion of attention information from multiple preset time windows. Attention information from a single preset time window is transient and susceptible to noise interference, making it difficult to accurately reflect the driver's true intentions and continuous behavioral patterns. This embodiment constructs a more representative periodic attention region by performing spatial union or intersection operations on the attention regions of multiple preset time windows, and dynamically adjusts the region brightness using a time decay mechanism, effectively reducing the risk of misjudgment caused by short-term fluctuations. This improves the stability and continuity of driver attention state recognition, providing a reliable basis for the formulation of subsequent object prompting strategies.

[0126] In this embodiment, the brightness of the region corresponding to the preset time period is determined based on the current brightness corresponding to each preset time window, including: for each sub-region in the attention region corresponding to the preset time period, the brightness of the sub-region is determined based on the current brightness corresponding to the preset time window to which the sub-region belongs; and the brightness of each sub-region is determined as the brightness of the region corresponding to the preset time period.

[0127] Specifically, the attention region corresponding to the preset time period can include multiple sub-regions, which can be overlapping or non-overlapping. For each sub-region, the preset time window to which the sub-region belongs is determined. If the sub-region is non-overlapping, there is one preset time window; if the sub-region is overlapping, there are at least two preset time windows. The current brightness corresponding to the preset time window to which the sub-region belongs is determined. If the sub-region corresponds to multiple preset time windows (i.e., overlapping regions), multiple current brightness values ​​can be determined; if the sub-region corresponds to one preset time window (i.e., non-overlapping regions), only one current brightness value can be determined.

[0128] For each sub-region, the regional brightness is determined based on one or more current brightness values ​​corresponding to that sub-region. For example, for overlapping regions, the final regional brightness can be selected from multiple current brightness values, such as the maximum value, weighted average, or sum. For non-overlapping regions, the corresponding current brightness value can be used as the regional brightness of that non-overlapping region. For example, for overlapping regions, if two preset time windows of attention regions overlap, the current brightness values ​​of these two time windows can be added together to obtain the regional brightness of the overlapping region. A maximum brightness value can be preset. After obtaining the regional brightness of a sub-region, it is compared with the maximum brightness value to ensure that the regional brightness of the sub-region does not exceed the maximum brightness value, thereby avoiding excessively high regional brightness after multiple superpositions.

[0129] The brightness of each sub-region is determined, and the brightness of each sub-region is combined to form the brightness of the region corresponding to a preset time period. That is, the regional intensity of the entire preset time period is not a globally uniform scalar, but a composite structure composed of multiple local regional brightness values. This design breaks through the limitations of traditional single numerical representation and can more realistically reflect the spatial heterogeneity of driver attention.

[0130] The beneficial effect of this setting is that the regional brightness allocation mechanism based on sub-region granularity allows for precise spatial reproduction of the attention intensity of different preset time windows, which helps to enhance the response accuracy and personalized adaptation capability of the intelligent driving assistance system.

[0131] In this embodiment, the method further includes: acquiring the driver's state information; wherein the state information represents the driver's attention state; and adjusting the brightness of the area corresponding to a preset time period based on the driver's state information to obtain the adjusted brightness of the area corresponding to the preset time period.

[0132] Specifically, after obtaining the regional brightness of the entire preset time period, that is, the regional brightness of each sub-region, further adjustments can be made according to the driver's state. For example, when the driver is detected to be slightly drowsy, the determined regional brightness can be multiplied by a preset coefficient to obtain the adjusted regional brightness, and then step S505 can be executed based on the adjusted regional brightness.

[0133] Acquiring driver status information refers to collecting data reflecting the driver's physiological and behavioral state through an in-vehicle perception system to assess their current level of cognitive focus. This status information can include parameters such as eyelid closure frequency, blink duration, head posture deviation angle, facial muscle activity intensity, heart rate variability, and steering wheel micro-movement frequency. These parameters are acquired through cameras, infrared sensors, bioelectric sensors, or wearable devices installed in the vehicle. For example, if the driver's eye closure time exceeds 1.5 seconds in multiple consecutive frames, or if the head tilt angle is greater than 20 degrees and maintained for more than 3 seconds, it is considered a state of fatigue; when the driver's gaze deviates from the road ahead for an extended period and fluctuates irregularly, it indicates a state of distraction.

[0134] State information represents the driver's attention state; that is, the collected data can be transformed into quantifiable attention evaluation indicators. For example, a normalized state scoring system can be constructed to fuse data collected from multiple modalities into a comprehensive attention score S. [0, 1], where S=1 indicates high alertness, and S close to 0 indicates severe fatigue or absent-mindedness. This score is implemented using a weighted fusion algorithm, machine learning classifier, or deep neural network to dynamically characterize the driver's cognitive resource allocation ability.

[0135] Different state information can be pre-associated with different state coefficients. For example, when S≥0.7, the state coefficient is 1.0; when 0.4≤S<0.7, the state coefficient is 0.7. The final output adjusted area brightness is the area brightness corresponding to the predetermined preset time period multiplied by the state coefficient. If there are multiple sub-regions, the area brightness of each sub-region is multiplied by the state coefficient separately.

[0136] The beneficial effect of this setup is that it enables attention evaluation optimization based on real-time individual status feedback, dynamically calibrates the authenticity of attention distribution, and thus improves the reliability and personalization of subsequent object prompt decisions.

[0137] S505. If an object is detected in the attention area corresponding to a preset time period, a strategy for sending a warning message to the driver is determined based on the object type information and the brightness of the area corresponding to the preset time period; wherein, the warning message is used to remind the driver to pay attention to the object.

[0138] For example, this step can refer to step S103 above, and will not be repeated here.

[0139] This application provides a driver attention-based warning method. It acquires the driver's gaze information within a preset time window, and determines the corresponding attention region and its brightness based on this gaze information. In other words, it determines the area the driver focuses on within the preset time window and the degree of attention paid to that area, achieving dynamic recognition of driver attention. Attention information from multiple preset time windows is fused to obtain attention information corresponding to a preset time period. That is, by using attention regions and their brightness from multiple short time windows, a long-period attention region and its brightness are formed, improving the temporal continuity and stability of attention judgment. Object recognition is performed on the attention region corresponding to the preset time period to determine the object's type information. Based on the object's type information and the corresponding brightness of the region within the preset time period, a strategy for sending warning information to the driver is determined, reminding the driver to pay targeted attention to the object, improving the accuracy of object warnings, realizing a differentiated prompting mechanism, and enhancing the safety and comfort of human-computer interaction.

[0140] Figure 8 A flowchart illustrating a driver attention-based warning method provided in this application embodiment is shown below. Figure 8 As shown, this embodiment, based on the above embodiments, provides a detailed description of a warning method based on driver attention. The method includes:

[0141] S801. Obtain the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window; wherein, the attention information includes the attention area and the area brightness, the attention area represents the area that the driver is paying attention to, and the area brightness represents the degree of attention the driver pays to the attention area.

[0142] For example, this step can refer to step S101 above, and will not be repeated here.

[0143] S802. Determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within the preset time period.

[0144] For example, this step can refer to step S102 above, and will not be repeated here.

[0145] S803. If an object is detected in the attention region corresponding to a preset time period, then the attention coefficient corresponding to the object type information is determined according to the preset second association relationship; wherein, the preset second association relationship represents the association relationship between the type information and the attention coefficient, and the attention coefficient represents the degree of difficulty for the driver to perceive the object.

[0146] For example, object detection is performed on the external environment of the vehicle in real time or at regular intervals, especially detecting whether there are objects in the attention region corresponding to a preset time period. If it is determined that there are objects in the attention region corresponding to the preset time period, the attention coefficient corresponding to the object can be determined. The attention coefficient represents the ease with which the driver perceives the object, that is, whether the driver has noticed the object.

[0147] Object type information refers to the category of a target object identified through image recognition, radar detection, or multi-sensor fusion technology, such as pedestrians, non-motorized vehicles, static obstacles, animals, or other motor vehicles. This type information is typically generated by vehicle-mounted perception systems such as cameras, millimeter-wave radar, and lidar, which collect raw data and then output classification labels and confidence scores through target detection algorithms. In this embodiment, no specific limitation is made on the target detection algorithm.

[0148] Attention coefficient is a quantitative indicator used to reflect how easily a driver naturally notices a specific type of object in a driving environment. A higher coefficient indicates that the object is more difficult for the driver to actively perceive, meaning it is more difficult to perceive; conversely, a lower coefficient indicates that the object is more easily noticed. For example, children quickly crossing the road are easily overlooked in complex backgrounds due to their small size and sudden movements, thus receiving a higher attention coefficient; while large trucks, due to their large size and clear outlines, are more easily noticed even at the periphery of the driver's field of vision, hence their lower attention coefficient. The attention coefficient can be normalized to a range of 0-1 or expressed as an integer, facilitating subsequent logical judgments and strategy matching.

[0149] The pre-defined second association is a pre-established mapping table or function model used to map object type information to corresponding attention coefficients. This association can be obtained through real-vehicle testing, simulated driving experiments, or analysis of historical accident data.

[0150] S804. Based on the attention coefficient corresponding to the object type information and the area brightness corresponding to the preset time period, determine the strategy for sending warning information to the driver.

[0151] For example, sending a warning message to the driver refers to outputting a warning signal through the in-vehicle human-machine interaction system, in forms including but not limited to audible alarms, visual cues, or tactile feedback. The content of the warning message should clearly indicate the presence of a potentially dangerous object that has not been adequately noticed, and suggest taking observation or avoidance actions.

[0152] The system compares the relative magnitude of the attention coefficient with the brightness of the attention area within a preset time period. The preset time period corresponds to multiple brightness levels within the attention area, and the maximum, minimum, average, and sum values ​​can be calculated. These calculated values ​​are then compared to the attention coefficient. For example, the attention coefficient can be compared to the maximum brightness level within the attention area of ​​the preset time period. If the object's attention coefficient is higher than the calculated brightness level, it indicates that the object is a highly concealed target, requiring a strong alert mechanism. Conversely, if the attention coefficient is lower than the calculated brightness level, it indicates that the object is relatively conspicuous and has already received sufficient attention; a weaker alert or no alert can be used to avoid interference.

[0153] This embodiment implements a dual assessment mechanism based on the perceived difficulty of object type and the driver's actual attention. By introducing an attention coefficient, a parameter that represents the perceived difficulty, and comparing it with the driver's actual attention intensity in the corresponding area, differentiated and accurate risk warnings are achieved, improving the usability and safety of advanced driver assistance systems, especially enhancing the active protection capability against highly concealed hazards.

[0154] In this embodiment, the attention region corresponding to the preset time period includes at least one sub-region; the strategy for sending warning information to the driver is determined based on the attention coefficient corresponding to the object type information and the brightness of the region corresponding to the preset time period, including: determining the sub-region to which the object belongs in the attention region corresponding to the preset time period as the target region; and determining the strategy for sending warning information to the driver based on the attention coefficient corresponding to the object type information and the brightness of the target region.

[0155] Specifically, the object's position information can be determined in real time. When the object is within the attention area, it can be further identified as belonging to a specific sub-region, which is then designated as the target region. The brightness of the target region can be obtained by considering the brightness of each sub-region within a preset time period.

[0156] The system compares the relative magnitudes of the attention coefficient and the brightness of the target area to intelligently decide whether to trigger a prompt and what intensity of prompt strategy to employ. For example, if the attention coefficient of an object is higher than the brightness of the target area, it indicates that the object is a highly concealed target. Although the driver may have scanned the area, the object's inherent characteristics prevented it from being effectively identified, thus requiring a strong alert mechanism. Conversely, if the attention coefficient is lower than the brightness of the target area, it indicates that the object is relatively conspicuous and has already received sufficient attention, allowing for a weaker prompt or no prompt at all to avoid interference.

[0157] The advantage of this setup is that it accurately locates the specific sub-region where the object is located, and combines the actual attention intensity of that sub-region with the object's perceived difficulty to make a comprehensive decision on whether to issue a prompt and the manner of the prompt. This avoids the false alarms or missed alarms caused by using global brightness as a threshold, thus improving the intelligence level of the driver assistance system and the efficiency of human-machine collaboration.

[0158] In this embodiment, a strategy for sending warning information to the driver is determined based on the attention coefficient corresponding to the object type information and the regional brightness of the target area. This includes: if the attention coefficient corresponding to the object type information is equal to or greater than the regional brightness of the target area, then a warning information is determined to be sent to the driver. The method further includes: sending a first warning information to the driver based on a preset first broadcast strategy.

[0159] Specifically, targeted warning messages can be issued based on different broadcast strategies. The content of the first and second warning messages can be the same or different. Multiple broadcast strategies are pre-set; for example, a first broadcast strategy and a second broadcast strategy can be included. The first broadcast strategy is used to issue the first warning message, and the second broadcast strategy is used to issue the second warning message. The first broadcast strategy is a high-intensity information reminder method, suitable for scenarios with high attention coefficients but low area brightness, meaning that there are potentially high-risk targets that have not been fully noticed. The first warning message corresponding to this strategy may include, but is not limited to: voice alarms (such as "Pedestrian ahead, please be careful!"), flashing red icons on the dashboard, steering wheel vibration, seat haptic feedback, etc. Such prompts have strong wake-up capabilities, aiming to quickly attract the driver's attention and prevent missed detection of key targets. As an alternative implementation, the intensity of the first broadcast strategy can be adaptively adjusted according to driving information such as vehicle speed and ambient light. For example, multimodal collaborative prompts (sound + vibration) can be added when driving at high speeds, while only visual prompts are used in low-speed urban environments to reduce interference.

[0160] The second alert strategy is a low-intensity or non-intrusive reminder method, suitable for targets with low attention thresholds or those already receiving sufficient attention, reflecting the comfort design principles of intelligent systems. The corresponding second warning information may include gentle voice prompts (such as "Right-side road tree identified"), green static icon displays, subtle sound effects, or no prompt at all. In some embodiments, the second alert strategy can also manifest as an information recording mode, storing relevant object information in the vehicle log for later analysis without disturbing the driver in real time. This strategy helps avoid information overload and maintain a good human-machine interaction experience.

[0161] When the attention coefficient of an object is determined to be greater than or equal to the brightness of the target area it occupies, the object is considered to be in a state of "should be noticed but is not currently being sufficiently noticed," triggering the first broadcast strategy; otherwise, the object is considered to have received sufficient attention or to have a low risk, and the second broadcast strategy is adopted. This differentiated response mechanism enables dynamic hierarchical control of cue intensity.

[0162] The warning broadcast function in this embodiment can be used in AEB (Autonomous Emergency Braking) and can also be applied in ADAS.

[0163] The beneficial effect of this setting is that it achieves a matching relationship between the cognitive difficulty of object type and the actual attention level, and intelligently selects the appropriate prompting strategy, which solves the problems of frequent false alarms or serious missed alarms, thus ensuring driving safety and improving the comfort of human-computer interaction.

[0164] This application provides a driver attention-based warning method. It acquires the driver's gaze information within a preset time window, and determines the corresponding attention region and its brightness based on this gaze information. In other words, it determines the area the driver focuses on within the preset time window and the degree of attention paid to that area, achieving dynamic recognition of driver attention. Attention information from multiple preset time windows is fused to obtain attention information corresponding to a preset time period. That is, by using attention regions and their brightness from multiple short time windows, a long-period attention region and its brightness are formed, improving the temporal continuity and stability of attention judgment. Object recognition is performed on the attention region corresponding to the preset time period to determine the object's type information. Based on the object's type information and the corresponding brightness of the region within the preset time period, a strategy for sending warning information to the driver is determined, reminding the driver to pay targeted attention to the object, improving the accuracy of object warnings, realizing a differentiated prompting mechanism, and enhancing the safety and comfort of human-computer interaction.

[0165] Figure 9 A schematic diagram of a driver attention-based warning device provided in this application embodiment is shown below. Figure 9 As shown, the driver attention-based warning device 90 provided in this embodiment includes:

[0166] The first determining unit 901 is used to acquire the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window; wherein, the attention information includes an attention area and an area brightness, the attention area represents the area that the driver is paying attention to, and the area brightness represents the degree of attention the driver pays to the attention area;

[0167] The second determining unit 902 is used to determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within the preset time period; and

[0168] The object detection unit 903 is used to determine a strategy for sending warning information to the driver based on the object type information and the brightness of the area corresponding to the preset time period if an object is detected in the attention area corresponding to the preset time period; wherein, the warning information is used to remind the driver to pay attention to the object.

[0169] In one possible implementation, the first determining unit 901 includes:

[0170] The feature determination module is used to determine the gaze features corresponding to the preset time window based on the gaze information within the preset time window; wherein, the gaze features characterize the driver's main gaze direction and the degree of dispersion of the gaze within the preset time window;

[0171] The information determination module is used to determine the attention information corresponding to the preset time window based on the gaze characteristics corresponding to the preset time window.

[0172] In one possible implementation, the gaze information within a preset time window includes the direction of the gaze at multiple moments within the preset time window; the feature determination module is specifically used for:

[0173] The gaze information within a preset time window is averaged to obtain the first feature corresponding to the preset time window; wherein, the first feature represents the driver's main gaze direction within the preset time window;

[0174] Variance processing is performed on the line-of-sight information within a preset time window to obtain a second feature corresponding to the preset time window; wherein, the second feature characterizes the degree of dispersion of the driver's line of sight within the preset time window;

[0175] Based on the first and second features, the gaze characteristics are determined.

[0176] In one possible implementation, the gaze characteristics include a first feature and a second feature. The first feature characterizes the driver's primary gaze direction within a preset time window, and the second feature characterizes the dispersion of the driver's gaze within the preset time window. The information determination module is specifically used for:

[0177] Based on the first and second features corresponding to the preset time window, determine the attention region corresponding to the preset time window;

[0178] The brightness of the region corresponding to the preset time window is determined based on the second feature corresponding to the preset time window.

[0179] In one possible implementation, the information determination module is specifically used for:

[0180] Based on the first feature corresponding to the preset time window, the region orientation information is determined; wherein, the region orientation information represents the fan-shaped opening direction of the attention region;

[0181] Based on the second feature corresponding to the preset time window, the central angle information is determined; wherein, the central angle information represents the fan-shaped opening range of the attention region;

[0182] The radius information is determined based on the line-of-sight information within a preset time window; the radius information represents the fan-shaped radius of the attention region.

[0183] Based on the region direction information, central angle information, and radius information, the attention region corresponding to the preset time window is determined.

[0184] In one possible implementation, the information determination module is specifically used for:

[0185] Based on the line-of-sight information within a preset time window, the line-of-sight depth value is determined; whereby the line-of-sight depth value represents the extended distance of the driver's line of sight.

[0186] The radius information is determined based on the line-of-sight depth value and a preset depth threshold.

[0187] One possible implementation also includes:

[0188] The information acquisition unit is used to acquire current driving information; the driving information includes at least vehicle speed information and environmental information.

[0189] The threshold determination unit is used to determine a threshold corresponding to the current driving information based on a preset first association relationship, which is a preset depth threshold; wherein, the preset first association relationship represents the association relationship between the driving information and the threshold.

[0190] In one possible implementation, the second determining unit 902 includes:

[0191] The region determination module is used to determine the attention region corresponding to the preset time period based on the attention region corresponding to each preset time window.

[0192] The brightness adjustment module is used to adjust the brightness of the area corresponding to each preset time window within a preset time period to obtain the current brightness corresponding to the preset time window.

[0193] The brightness determination module is used to determine the regional brightness corresponding to a preset time period based on the current brightness of each preset time window.

[0194] In one possible implementation, the region determination module is specifically used for:

[0195] Based on the attention regions corresponding to each preset time window, at least one sub-region is determined; wherein, the sub-region includes overlapping regions and / or non-overlapping regions, the overlapping region represents the intersection of the attention regions between at least two preset time windows, and the non-overlapping region represents the region in the attention region corresponding to the preset time window other than the overlapping region.

[0196] At least one sub-region is identified as the attention region corresponding to a preset time period.

[0197] In one possible implementation, the brightness adjustment module is specifically used for:

[0198] Based on the preset brightness attenuation relationship information, the brightness attenuation coefficient of the preset time window is determined; whereby the brightness attenuation relationship information represents the relationship between the brightness attenuation coefficient and time, and the brightness attenuation coefficient represents the degree of brightness attenuation in the region.

[0199] Based on the brightness attenuation coefficient of the preset time window, the brightness of the area corresponding to the preset time window is adjusted to obtain the current brightness corresponding to the preset time window.

[0200] In one possible implementation, the brightness determination module is specifically used for:

[0201] For each sub-region within the attention region corresponding to the preset time period, the region brightness of the sub-region is determined based on the current brightness of the preset time window to which the sub-region belongs.

[0202] The brightness of each sub-region is determined to be the brightness of the region corresponding to a preset time period.

[0203] One possible implementation also includes:

[0204] A state acquisition unit is used to acquire the driver's state information; wherein, the state information represents the driver's attention state;

[0205] The status adjustment unit is used to adjust the brightness of the area corresponding to a preset time period based on the driver's status information, so as to obtain the adjusted brightness of the area corresponding to the preset time period.

[0206] In one possible implementation, the object detection unit 903 includes:

[0207] The coefficient determination module is used to determine the attention coefficient corresponding to the object type information based on the preset second association relationship; wherein, the preset second association relationship represents the association relationship between the type information and the attention coefficient, and the attention coefficient represents the degree of difficulty for the driver to perceive the object;

[0208] The warning module is used to determine the strategy for sending warning information to the driver based on the attention coefficient corresponding to the type information of the object and the area brightness corresponding to the preset time period.

[0209] In one possible implementation, the attention region corresponding to the preset time period includes at least one sub-region; the early warning module is specifically used for:

[0210] The sub-region to which the object belongs in the attention area corresponding to the preset time period is identified as the target region;

[0211] Based on the attention coefficient corresponding to the object type information and the regional brightness of the target area, a strategy for sending warning information to the driver is determined.

[0212] In one possible implementation, the early warning module is specifically used for:

[0213] If the attention coefficient corresponding to the object type information is equal to or greater than the regional brightness of the target area, then a warning message is sent to the driver.

[0214] The device also includes a first warning unit, used to send a first warning message to the driver based on a preset first broadcast strategy.

[0215] This embodiment provides a warning device based on driver attention, which can execute the method provided in the above-described method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0216] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 10 As shown, the electronic device 1000 provided in this embodiment includes at least one processor 1001 and a memory 1002. Optionally, the device 1000 further includes a communication component 1003. The processor 1001, memory 1002, and communication component 1003 are connected via a bus 1004.

[0217] In a specific implementation, at least one processor 1001 executes computer execution instructions stored in memory 1002, causing at least one processor 1001 to perform the above-described method.

[0218] The specific implementation process of processor 1001 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0219] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0220] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0221] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0222] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0223] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0224] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0225] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0226] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0227] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0228] In addition, the functional units in the various embodiments of the present invention 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.

[0229] If a function is implemented as a software functional unit and 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 invention, or the part that contributes to the prior art, or a 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.) to execute all or part of the steps of the methods of the various embodiments of this invention. 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.

[0230] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0231] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A warning method based on driver attention, characterized in that, include: Acquire the driver's line-of-sight information within a preset time window; Based on the gaze information within the preset time window, the attention information corresponding to the preset time window is determined, wherein the attention information includes an attention region and a region brightness, the attention region represents the area that the driver is focused on, and the region brightness represents the degree of attention the driver pays to the attention region; Based on the attention information corresponding to multiple preset time windows within a preset time period, determine the attention information corresponding to the preset time period; and If an object is detected in the attention area corresponding to the preset time period, a strategy for sending a warning message to the driver is determined based on the object's type information and the brightness of the area corresponding to the preset time period. The warning message is used to remind the driver to pay attention to the object.

2. The method according to claim 1, characterized in that, Based on the gaze information within the preset time window, determine the attention information corresponding to the preset time window, including: Based on the gaze information within the preset time window, the gaze characteristics corresponding to the preset time window are determined; wherein, the gaze characteristics characterize the driver's main gaze direction and the degree of gaze dispersion within the preset time window; Based on the gaze characteristics corresponding to the preset time window, determine the attention information corresponding to the preset time window.

3. The method according to claim 2, characterized in that, The line-of-sight information within the preset time window includes the direction of the line of sight at multiple moments within the preset time window; Based on the gaze information within the preset time window, determine the gaze characteristics corresponding to the preset time window, including: The gaze information within the preset time window is averaged to obtain a first feature corresponding to the preset time window; wherein, the first feature represents the driver's main gaze direction within the preset time window. Variance processing is performed on the line-of-sight information within the preset time window to obtain a second feature corresponding to the preset time window; wherein, the second feature characterizes the degree of dispersion of the driver's line of sight within the preset time window; The gaze feature is determined based on the first feature and the second feature.

4. The method according to claim 2, characterized in that, The gaze characteristics include a first feature and a second feature. The first feature represents the driver's main gaze direction within a preset time window, and the second feature represents the degree of dispersion of the driver's gaze within the preset time window. Based on the gaze characteristics corresponding to the preset time window, the attention information corresponding to the preset time window is determined, including: Based on the first feature and the second feature corresponding to the preset time window, the attention region corresponding to the preset time window is determined; The brightness of the region corresponding to the preset time window is determined based on the second feature corresponding to the preset time window.

5. The method according to claim 4, characterized in that, Based on the first and second features corresponding to the preset time window, the attention region corresponding to the preset time window is determined, including: Based on the first feature corresponding to the preset time window, the region direction information is determined, wherein the region direction information represents the fan-shaped opening direction of the attention region; Based on the second feature corresponding to the preset time window, the central angle information is determined, wherein the central angle information represents the fan-shaped opening range of the attention region; Based on the line-of-sight information within the preset time window, radius information is determined, wherein the radius information represents the fan-shaped radius of the attention region; Based on the region direction information, the central angle information, and the radius information, the attention region corresponding to the preset time window is determined.

6. The method according to claim 5, characterized in that, Based on the line-of-sight information within the preset time window, the radius information is determined, including: Based on the line-of-sight information within the preset time window, a line-of-sight depth value is determined, wherein the line-of-sight depth value represents the extension distance of the driver's line of sight; The radius information is determined based on the line-of-sight depth value and a preset depth threshold.

7. The method according to claim 6, characterized in that, Also includes: Obtain current driving information, wherein the driving information includes at least vehicle speed information and environmental information; Based on a preset first association relationship, a threshold corresponding to the current driving information is determined, which is the preset depth threshold, wherein the preset first association relationship characterizes the association relationship between the driving information and the threshold.

8. The method according to claim 1, characterized in that, Based on the attention information corresponding to multiple preset time windows within a preset time period, the attention information corresponding to the preset time period is determined, including: Based on the attention regions corresponding to each preset time window, determine the attention region corresponding to the preset time period; For each preset time window within the preset time period, the brightness of the area corresponding to the preset time window is adjusted to obtain the current brightness corresponding to the preset time window; The brightness of the region corresponding to the preset time period is determined based on the current brightness corresponding to each preset time window.

9. The method according to claim 8, characterized in that, Based on the attention regions corresponding to each preset time window, the attention region corresponding to the preset time period is determined, including: Based on the attention regions corresponding to each preset time window, at least one sub-region is determined, wherein the sub-region includes overlapping regions and / or non-overlapping regions, the overlapping regions represent the intersection of the attention regions between at least two preset time windows, and the non-overlapping regions represent the regions in the attention regions corresponding to the preset time windows other than the overlapping regions; The at least one sub-region is determined as the attention region corresponding to the preset time period.

10. The method according to claim 8, characterized in that, Adjusting the brightness of the area corresponding to the preset time window to obtain the current brightness corresponding to the preset time window includes: Based on preset brightness attenuation relationship information, the brightness attenuation coefficient of the preset time window is determined, wherein the brightness attenuation relationship information characterizes the relationship between the brightness attenuation coefficient and time, and the brightness attenuation coefficient characterizes the degree of brightness attenuation in the region. The brightness of the area corresponding to the preset time window is adjusted according to the brightness attenuation coefficient of the preset time window to obtain the current brightness corresponding to the preset time window.

11. The method according to claim 9, characterized in that, Based on the current brightness corresponding to each preset time window, determine the regional brightness corresponding to the preset time period, including: For each sub-region in the attention region corresponding to the preset time period, the region brightness of the sub-region is determined according to the current brightness of the preset time window to which the sub-region belongs; The brightness of each sub-region is determined to be the brightness of the region corresponding to the preset time period.

12. The method according to claim 8, characterized in that, Also includes: Acquire driver status information; wherein the status information represents the driver's attention state; Based on the driver's status information, the brightness of the area corresponding to the preset time period is adjusted to obtain the adjusted brightness of the area corresponding to the preset time period.

13. The method according to any one of claims 1-12, characterized in that, Based on the object type information and the regional brightness corresponding to the preset time period, a strategy for sending warning information to the driver is determined, including: Based on a preset second association relationship, an attention coefficient corresponding to the type information of the object is determined, wherein the preset second association relationship represents the association relationship between the type information and the attention coefficient, and the attention coefficient represents the degree of difficulty for the driver to perceive the object; Based on the attention coefficient corresponding to the type information of the object and the regional brightness corresponding to the preset time period, a strategy for sending warning information to the driver is determined.

14. The method according to claim 13, characterized in that, The attention region corresponding to the preset time period includes at least one sub-region; based on the attention coefficient corresponding to the object type information and the brightness of the region corresponding to the preset time period, a strategy for sending warning information to the driver is determined, including: The sub-region to which the object belongs in the attention region corresponding to a preset time period is determined as the target region; Based on the attention coefficient corresponding to the type information of the object and the regional brightness of the target area, a strategy for sending warning information to the driver is determined.

15. The method according to claim 14, characterized in that, Based on the attention coefficient corresponding to the type information of the object and the regional brightness of the target area, a strategy for sending warning information to the driver is determined, including: If the attention coefficient corresponding to the type information of the object is equal to or greater than the regional brightness of the target area, then it is determined to send a warning message to the driver. The method further includes: sending a first warning message to the driver based on a preset first broadcast strategy.

16. A warning device based on driver attention, characterized in that, include: The first determining unit is used to acquire the driver's gaze information within a preset time window, and determine the attention information corresponding to the preset time window based on the gaze information within the preset time window. The attention information includes an attention region and a region brightness. The attention region represents the area that the driver is paying attention to, and the region brightness represents the degree of attention the driver pays to the attention region. The second determining unit is configured to determine the attention information corresponding to the preset time period based on the attention information corresponding to multiple preset time windows within a preset time period; and An object detection unit is configured to determine a strategy for sending a warning message to the driver based on the object's type information and the brightness of the area corresponding to the preset time period if an object is detected in the attention area corresponding to the preset time period. The warning message is used to remind the driver to pay attention to the object.

17. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-15.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-15.

19. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-15.

Images