Eye lid closure detection method, alert method, device, vehicle, medium and apparatus

Abstract

The application discloses an eyelid closure detection method, an alarm method, a device, a vehicle, a medium and equipment, and belongs to the field of data processing. The method comprises the following steps: a sensing model performs sensing detection on the positions of the binocular reference points of personnel in each current image, distance calculation is performed on the positions of the binocular reference points, current eye width is obtained, and the current eye width is mapped into a normalized interval to obtain a current width normalized value; in a preset time period, a plurality of current width normalized values obtained are sorted to obtain a width time sequence queue, and weighted calculation is performed on the current width normalized values in the width time sequence queue to obtain a width average value of the width time sequence queue; and the width average value is compared with a width threshold value to determine that the eye state of the personnel is eyelid closure when the width average value is less than the width threshold value. The application improves the accuracy of personnel eyelid closure detection, reduces the false positive rate, and improves the user experience.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to an eyelid closure detection method, alarm method, device, vehicle, medium and equipment. Background Technology

[0002] The detection of eyelid closure status is being used more and more widely in various industries. For example, eyelid closure detection for drivers is mainly used to detect driver fatigue; eyelid closure detection for students taking online classes is mainly used to detect whether students are dozing off during class; eyelid closure detection for sales clerks is mainly used to detect whether sales clerks are dozing off during work, and so on.

[0003] However, because different people have different eye sizes and different eye opening distances, under the same judgment standard, the eyelid test results for people with small eyes are often inaccurate. Summary of the Invention

[0004] To address the issue of false detections caused by differences in eye size among individuals, this application primarily provides an eyelid closure detection method, alarm method, device, vehicle, medium, and equipment.

[0005] In a first aspect, embodiments of this application provide a method for detecting eyelid closure, comprising:

[0006] Perception detection is performed on people in a predetermined number of historical images, and the positions of binocular reference points in the predetermined number of historical images are marked, where the binocular reference points include the left eye reference point and the right eye reference point.

[0007] Based on the position of the reference points of both eyes, the distance between the upper and lower eyelids of each eye in a predetermined number of historical images is calculated to obtain the eye width in each historical image. The eye widths in the predetermined number of historical images are then filtered to obtain the maximum eye width. Finally, the maximum eye width is normalized to obtain the normalized interval.

[0008] The perception model calculates the distance to the positions of the binocular reference points detected by the perception model for the person in each frame of the current image, obtains the current eye width, and maps the current eye width to a normalized interval to obtain the current width normalized value.

[0009] Within a preset time period, the obtained multiple current width normalized values ​​are sorted to obtain a width time series queue, and the current width normalized values ​​in the width time series queue are weighted to obtain the average width of the width time series queue.

[0010] The average width is compared with the width threshold. When the average width is less than the width threshold, the person's eye state is determined to be closed eyelids.

[0011] Secondly, embodiments of this application provide a driver's eyelid closure alarm method, which includes:

[0012] The driver's eyelid closure status is detected using the method in Scheme 2; and an alarm is issued to the driver when the driver's eyelids are detected to be closed.

[0013] Thirdly, embodiments of this application provide an eyelid closure detection device, which includes:

[0014] The perception information acquisition module is used to perform perception detection on people in a predetermined number of historical images and to mark the positions of binocular reference points in the predetermined number of historical images, wherein the binocular reference points include a left eye reference point and a right eye reference point.

[0015] The normalization calculation module is used to calculate the distance between the upper and lower eyelids of each eye in a predetermined number of historical images based on the position of the binocular reference points, to obtain the eye width in each historical image, and to filter the eye widths in the predetermined number of historical images to obtain the maximum eye width, and to normalize the maximum eye width to obtain the normalization interval.

[0016] The width normalization module is used to calculate the distance between the positions of the binocular reference points detected by the perception model for the person in each frame of the current image, to obtain the current eye width, and to map the current eye width into the normalization interval to obtain the current width normalization value.

[0017] The width calculation module is used to sort multiple current width normalized values ​​within a preset time period to obtain a width time sequence queue, and to perform weighted calculation on the current width normalized values ​​in the width time sequence queue to obtain the average width of the width time sequence queue.

[0018] The behavior determination module is used to compare the average width with the width threshold. When the average width is less than the width threshold, the person's eye state is determined to be closed eyelids.

[0019] Fourthly, embodiments of this application provide a vehicle, wherein the vehicle includes the eyelid closure detection device of embodiment three.

[0020] Fifthly, embodiments of this application provide a computer-readable storage medium storing computer instructions. When the computer instructions are executed, the computer performs the eyelid closure detection method in Scheme 1 or the driver eyelid closure alarm method in Scheme 2.

[0021] Sixthly, embodiments of this application provide a computer device, which includes a processor and a memory. The memory stores computer instructions, and when the computer instructions are executed by the processor, they implement the eyelid closure detection method in Scheme 1 or the driver eyelid closure alarm method in Scheme 2.

[0022] The technical solution of this application calculates and filters the eye width of each eye in a predetermined number of historical image frames to obtain the maximum eye width. A normalized range is then obtained through a normalization magnification operation. Next, the current eye width detected by the perception model in each current image frame is normalized and mapped to obtain a normalized current width value. This value is then sorted and weighted to calculate the average width. The average width is compared with a width threshold; an eyelid closure alarm is only triggered if the average width within a preset time period is less than the threshold. This application improves the accuracy of eyelid closure detection, reduces false alarms, and enhances the user experience by setting a normalized range. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description exemplarily illustrate some embodiments of this application.

[0024] Figure 1 This is a schematic diagram illustrating a specific implementation of an eyelid closure detection method according to this application;

[0025] Figure 2 This is a schematic diagram of a specific embodiment of an eyelid closure detection device according to this application.

[0026] 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

[0027] The preferred embodiments of this application will now be described in detail with reference to the accompanying drawings, so that the advantages and features of this application can be more easily understood by those skilled in the art, thereby providing a clearer and more definite definition of the scope of protection of this application.

[0028] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0029] The technical solutions of this application and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. The specific embodiments described below can be combined with each other to form new embodiments. The same or similar ideas or processes described in one embodiment may not be repeated in other embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0030] The eyelid closure detection method of this application is applicable to various scenarios that require the detection of the degree of eye closure, such as fatigue driving detection scenarios, student drowsiness detection scenarios, salesperson drowsiness detection scenarios, etc. As long as the scenario involves detecting the eyelid closure status of a person, the technical solution of this application can be adopted. The following will use the detection of the driver's eyelid closure status in a fatigue driving detection scenario as an example to illustrate the technical solution of this application.

[0031] Figure 1 This paper illustrates a specific embodiment of an eyelid closure detection method according to the present application.

[0032] exist Figure 1 In the specific embodiments shown, an eyelid closure detection method mainly includes:

[0033] Step S101: Perform perception detection on the people in a predetermined number of historical images and mark the positions of the binocular reference points in the predetermined number of historical images, wherein the binocular reference points include the left eye reference point and the right eye reference point.

[0034] In this embodiment, in the three-dimensional camera coordinate system corresponding to the vehicle camera parameters, in a predetermined number of historical images, the perception model detects the eye position coordinates of the person in each frame of the two-dimensional image coordinate system, such as the driver, in the three-dimensional camera coordinate system, laying the foundation for subsequent eyelid closure judgment.

[0035] In one specific embodiment of this application, the facial key point detection results of a person output by the perception model are used to determine whether the eyes of the person in each frame image are occluded. When the facial key point detection result indicates that the eye key points are missing, the frame image is filtered out.

[0036] In this embodiment, the perception model obtains the position of the reference points of the eyes of people in the image, such as the driver, based on the position coordinates of the detected facial key points. Therefore, the perception model also outputs the facial key point detection results, which include which key points are missing. If the eye key points are missing, regardless of whether the number of facial key points is less than the number threshold, it is considered that the eyes of the people in the image, such as the driver, are obstructed, indicating that the frame of the image needs to be discarded to prevent interference to subsequent steps due to incomplete eye key points.

[0037] exist Figure 1 In the specific embodiment shown, an eyelid closure detection method further includes:

[0038] Step S102: Based on the position of the reference points of both eyes, calculate the distance between the upper and lower eyelids of each eye in a predetermined number of historical images to obtain the eye width in each historical image. Then, filter the eye widths in the predetermined number of historical images to obtain the maximum eye width and normalize the maximum eye width to obtain the normalized interval.

[0039] In this embodiment, the width of each eye in each frame of historical images is calculated, i.e., the distance between the upper and lower eyelids is calculated. This distance is at least 0, so the minimum eye width is also 0. By comparing and filtering the eye widths of the predetermined number of historical images, the eye width with the largest eye opening distance in the selected image is taken as the maximum eye width. The maximum eye width is then normalized and enlarged. The minimum eye width and the maximum eye width are used to obtain the normalization interval. The setting of the normalization interval makes the judgment of eyelid closure more accurate and reduces the false judgment rate.

[0040] In one specific embodiment of this application, the distance between the upper and lower eyelids of each eye in a predetermined number of historical images is calculated based on the positions of the binocular reference points to obtain the eye width in each historical image. This includes: calculating the width of the left eye and the right eye in each historical image based on the positions of the left eye reference point and the right eye reference point in the binocular reference points to obtain the eye width corresponding to each historical image, wherein the eye width includes the left eye width and the right eye width.

[0041] In this embodiment, there are two eyes in each frame of historical image, and there are four or more reference point coordinates for each eye, including the coordinates of the two corners of the eye, the coordinates of the upper eyelid and the lower eyelid; the eye width of each eye can be obtained based on the coordinates of the upper and lower eyelids of each eye, and this calculation method can improve the accuracy.

[0042] In one specific embodiment of this application, the maximum eye width is obtained by filtering the eye width in a predetermined number of historical images, including: comparing the left eye width and right eye width in each historical image frame in the predetermined number of historical images to filter out the maximum eye width in each historical image frame; and comparing the maximum eye width in the predetermined number of historical images to filter out the maximum eye width in one historical image frame.

[0043] In this embodiment, each frame of historical image contains two eye widths: the left eye width and the right eye width. Since some drivers have eyes of different sizes, it is necessary to compare the maximum widths of the left and right eyes in each frame of historical image to determine the eye width with the largest distance in that frame, i.e., the maximum eye width. By filtering out the minimum width between the two eyes in each frame of historical image, the accuracy of the driver fatigue monitoring system can be improved.

[0044] In one specific embodiment of this application, the maximum width of the eye is normalized to obtain a normalized interval, including: mapping the maximum width of the eye to the normalized maximum value through a magnification operation, as the upper limit of the normalized interval; and mapping the width when the eyes are closed to the normalized minimum value, as the lower limit of the normalized interval.

[0045] In this embodiment, the maximum eye width is mapped to the normalized maximum value, and the eye width when the eyes are closed is mapped to the normalized minimum value. The resulting normalization interval amplifies the width of the eyes in each frame, which plays an important role in reducing the false alarm rate.

[0046] In a specific example of this application, assuming the maximum eye width is 0.5 cm, the maximum eye width of 0.5 cm is mapped to a normalized maximum value of 100, and the minimum eye width is 0, which is mapped to a normalized minimum value of 0. The resulting normalization interval is 0 to 100. If the distance between the driver's eyes in the current frame image is 0.4 cm, 0.3 cm, or 0.2 cm, mapping this distance to the normalization interval yields normalized values ​​of 80, 60, or 40, respectively.

[0047] exist Figure 1 In the specific embodiment shown, an eyelid closure detection method further includes:

[0048] Step S103: The perception model calculates the distance to the positions of the reference points of the two eyes detected by the perception model in each frame of the current image to obtain the current eye width, and maps the current eye width to a normalized interval to obtain the current width normalized value.

[0049] In this embodiment, the perception model transforms the position of the binocular reference point in the current image under the camera coordinate system to the position of the binocular reference point in the two-dimensional image coordinate system, calculates the current eye width in each frame of the current image, maps the current eye width to a normalized interval, and enlarges the current eye width to make subsequent calculations more accurate.

[0050] exist Figure 1 In the specific embodiment shown, an eyelid closure detection method further includes:

[0051] Step S104: Within a preset time period, sort the multiple current width normalized values ​​to obtain a width time sequence queue, and perform weighted calculation on the current width normalized values ​​in the width time sequence queue to obtain the average width of the width time sequence queue.

[0052] In this embodiment, each current image frame is sorted according to its arrival order to obtain a width time sequence queue within a preset time period. The average width is calculated from the current normalized width values ​​in the width time sequence queue, providing a data basis for subsequent eyelid closure judgment. To ensure the accuracy of eyelid closure judgment, it is necessary to calculate the average value of the current normalized width values ​​in the width time sequence queue, i.e., the average width, which lays the foundation for subsequent judgment on whether the fatigue driving detection system should issue an alarm.

[0053] exist Figure 1 In the specific embodiment shown, an eyelid closure detection method further includes:

[0054] Step S105: Compare the average width with the width threshold. When the average width is less than the width threshold, determine that the person's eye state is closed.

[0055] In this embodiment, when the average width within a preset time period is less than the width threshold, it indicates that the person, such as the driver, has been continuously closing their eyes for a period of time. This confirms the person, such as the driver, whose eyelids are closed, and lays the foundation for whether the fatigue driving detection system should issue an alarm.

[0056] In another specific embodiment of this application, a driver eyelid closure alarm method includes: detecting the driver's eyelid closure state using the eyelid closure detection method in any embodiment; and issuing an alarm to the driver when the driver's eyelids are detected to be in a closed state.

[0057] In this embodiment, when it is detected that a person is indeed in a closed eyelid state, the person can be promptly alerted, and the fatigue driving system can also promptly remind the driver in dangerous situations.

[0058] In one specific embodiment of this application, when the speed of the vehicle driven by the person exceeds a speed threshold, the person's eyelid closure behavior is re-evaluated and an alarm is triggered every preset time interval.

[0059] In this embodiment, an alarm is triggered to detect dangerous eyelid closure by the driver during driving, provided the vehicle speed is greater than 0. Setting a preset alarm time interval ensures more timely and accurate alarms.

[0060] In a specific example of this application, the alarm for the dangerous behavior of eyelid closure during driving is not continuous, but rather re-alarms at preset time intervals. Because the danger of eyelid closure is relatively high, the corresponding alarm interval is set to 5 seconds.

[0061] This application, in the three-dimensional camera coordinate system corresponding to the vehicle-mounted camera parameters, sequentially performs perception detection on the driver in a predetermined number of historical images using a perception model, calibrating the positions of the driver's binocular reference points in the two-dimensional image coordinate system, laying the data foundation for subsequent eyelid closure judgment; based on the positions of the binocular reference points, the distance between the upper and lower eyelids of each eye in the predetermined number of historical images is calculated to obtain the eye width in each historical image, and the eye widths in the predetermined number of historical images are filtered to obtain the maximum eye width, and the maximum eye width is normalized to obtain a normalization interval. The setting of the normalization interval makes the judgment of eyelid closure more accurate and reduces the false judgment rate; the position of the binocular reference points detected by the perception model on the driver in each current image is then used to determine the driver's position. The system performs distance calculations to obtain the current eye width for each eye, and maps the current eye width to a normalized interval to obtain a normalized value. By mapping the current eye width to the normalized interval, the current eye width is magnified, making subsequent calculations more accurate. Within a preset time period, the obtained multiple normalized values ​​of the current width are sorted to obtain a width time sequence queue. The normalized values ​​of the current width in the width time sequence queue are then weighted to obtain the average width of the width time sequence queue, laying the foundation for subsequent judgment on whether the system should alarm. The average width is compared with a width threshold. When the average width is less than the width threshold, an alarm is triggered for the driver's eyelid closure behavior. This not only promptly alerts the driver to dangerous behavior but also improves the accuracy of driver eyelid closure detection.

[0062] Figure 2This paper illustrates a specific embodiment of an eyelid closure detection device according to the present application.

[0063] exist Figure 2 In the specific embodiment shown, an eyelid closure detection device mainly includes:

[0064] The perception information acquisition module 201 is used to perform perception detection on people in a predetermined number of historical images and to mark the positions of binocular reference points in the predetermined number of historical images, wherein the binocular reference points include a left eye reference point and a right eye reference point.

[0065] The normalization calculation module 202 is used to calculate the distance between the upper and lower eyelids of each eye in a predetermined number of historical images based on the position of the reference points of both eyes, to obtain the eye width in each historical image, and to filter the eye widths in the predetermined number of historical images to obtain the maximum eye width, and to normalize the maximum eye width to obtain the normalization interval.

[0066] The width normalization module 203 is used to calculate the distance between the positions of the binocular reference points detected by the perception model for the person in each frame of the current image, to obtain the current eye width, and to map the current eye width into the normalization interval to obtain the current width normalization value.

[0067] The width calculation module 204 is used to sort multiple current width normalized values ​​within a preset time period to obtain a width time sequence queue, and to perform weighted calculation on the current width normalized values ​​in the width time sequence queue to obtain the average width of the width time sequence queue.

[0068] The behavior determination module 205 is used to compare the average width with the width threshold. When the average width is less than the width threshold, the person's eye state is determined to be closed eyelids.

[0069] In this embodiment, the perception information acquisition module 201 detects and outputs the position coordinates of the driver's eyes in each frame of the image in the two-dimensional image coordinate system within a predetermined number of historical image frames, using a perception model in the three-dimensional camera coordinate system. The perception model also filters images based on facial key point detection results. These results include identifying which key points are missing. If eye key points are missing, regardless of whether the number of facial key points is less than a threshold, the driver's eyes are considered occluded, indicating that the frame needs to be discarded, thus providing a data basis for the successful implementation of this scheme. The normalization calculation module 202 calculates the width of each eye in each frame of the historical image within the predetermined number of historical image frames, specifically calculating the distance between the upper and lower eyelids. This distance is at least 0, therefore the minimum eye width is also 0. By comparing and filtering the eye widths of the predetermined number of historical image frames, the selected images are chosen. The maximum eye width is defined as the eye width at which the eye is widest when open. This maximum eye width is then normalized and amplified. A normalization interval is obtained by comparing the minimum eye width with the maximum eye width. This normalization interval makes the judgment of eyelid closure more accurate and reduces the false positive rate. The width normalization module 203 transforms the position of the binocular reference points in the current image in the camera coordinate system to the position of the binocular reference points in the two-dimensional image coordinate system. It calculates the current eye width in each frame of the current image and maps it to the normalization interval, amplifying the current eye width for more accurate subsequent calculations. The width calculation module 204 sorts each frame of the current image according to the order in which they arrive, obtaining a width time-series queue within a preset time period. The normalized width values ​​in the width time-series queue are calculated to obtain the average width, providing a data basis for subsequent eyelid closure alarm judgment. To ensure the accuracy of the eyelid closure alarm judgment, it is necessary to calculate the average value of the current normalized width of the width time sequence queue, i.e., the average width, to lay the foundation for subsequent judgment of whether the system should alarm; the behavior determination module 205, when the average width within a preset time period is less than the width threshold, indicates that the driver has been continuously closing his eyes for the current period of time, and alarms are issued for the driver's eyelid closure behavior, which can promptly remind the driver in dangerous situations.

[0070] The eyelid closure detection device provided in this application can be used to perform the eyelid closure detection method described in any of the above embodiments. Its implementation principle and technical effect are similar, and will not be repeated here.

[0071] In one specific embodiment of this application, the functional modules of the eyelid closure detection device proposed in this application can be directly in hardware, in software modules executed by a processor, or in a combination of both.

[0072] Software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in this art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium.

[0073] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor can be a microprocessor, but alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors incorporating a DSP core, or any other such configuration. Alternatively, the storage medium can be integrated with the processor. The processor and storage medium can reside in an ASIC. The ASIC can reside in the user terminal. Alternatively, the processor and storage medium can reside as discrete components in the user terminal.

[0074] In another specific embodiment of this application, a vehicle is provided, wherein the vehicle includes the eyelid closure detection device of any embodiment. Optionally, the vehicle includes a processor and a memory, the processor and memory being coupled, the vehicle being used to implement the appendix to this specification. Figure 1 The eyelid closure detection method in any of the embodiments shown.

[0075] In another specific embodiment of this application, a computer-readable storage medium is provided, which stores computer instructions that, when executed, cause a computer to perform the eyelid closure detection method or the driver eyelid closure alarm method in any embodiment.

[0076] In another specific embodiment of this application, a computer device is provided, which includes a processor and a memory. The memory stores computer instructions, which, when executed by the processor, implement the eyelid closure detection method or the driver eyelid closure alarm method in any embodiment.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0078] 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.

[0079] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for detecting eyelid closure, characterized in that, include: Perception detection is performed on people in a predetermined number of historical images, and the positions of binocular reference points in the predetermined number of historical images are marked, wherein the binocular reference points include a left eye reference point and a right eye reference point; Based on the positions of the binocular reference points, the distance between the upper and lower eyelids of each eye in the predetermined number of historical images is calculated to obtain the eye width in each frame of the historical image. The eye widths in the predetermined number of historical images are then filtered to obtain the maximum eye width. The maximum eye width is then normalized to obtain a normalized interval. The normalized interval is obtained by using the minimum eye width and the maximum eye width. The perception model calculates the distance to the positions of the binocular reference points detected by the perception model for the person in each frame of the current image, obtains the current eye width, and maps the current eye width to the normalization interval to obtain the current width normalization value. The maximum eye width is mapped to the normalized maximum value, and the eye width when the eyes are closed is mapped to the normalized minimum value, so that the current eye width is magnified when mapped to the normalization interval. Within a preset time period, the multiple obtained current width normalized values ​​are sorted to obtain a width time series queue, and the current width normalized values ​​in the width time series queue are weighted to obtain the average width of the width time series queue. The average width is compared with a width threshold. When the average width is less than the width threshold, the person's eye state is determined to be closed eyelids.

2. The eyelid closure detection method as described in claim 1, characterized in that, The step of calculating the distance between the upper and lower eyelids of each eye in the predetermined number of historical images based on the positions of the binocular reference points, to obtain the eye width in each frame of the historical image, includes: Based on the positions of the left eye reference point and the right eye reference point in the binocular reference points, the widths of the left and right eyes in each frame of the historical image are calculated to obtain the eye width corresponding to each frame of the historical image, wherein the eye width includes the left eye width and the right eye width.

3. The eyelid closure detection method as described in claim 2, characterized in that, The step of filtering the eye widths in the predetermined number of historical images to obtain the maximum eye width includes: In the predetermined number of historical images, the width of the left eye and the width of the right eye in each historical image are compared to filter out the maximum width of the eye in each historical image. The maximum width of the eye in the predetermined number of historical images is compared, and the maximum width of the eye in one frame of the historical image is selected.

4. The eyelid closure detection method as described in claim 1, characterized in that, Also includes: Using the facial key point detection results of the person output by the perception model, it is determined whether the person's eyes are occluded in each frame of the image. When the facial key point detection result indicates that eye key points are missing, the frame of the image is filtered out.

5. The eyelid closure detection method as described in claim 1, characterized in that, The normalization of the maximum width of the eye to obtain a normalized interval includes: The maximum width of the eye is mapped to the normalized maximum value through a magnification operation, which serves as the upper limit of the normalization interval; The width when the eyes are closed is mapped to the normalized minimum value, which serves as the lower limit of the normalized interval.

6. A method for triggering an alarm for driver's eyelid closure, characterized in that, include: The driver's eyelid closure status is detected by the method of any one of claims 1-5; as well as An alarm is issued to the driver if the driver's eyelids are detected to be closed.

7. The driver's eyelid closure alarm method as described in claim 6, characterized in that, Also includes: When the speed of the driver's vehicle exceeds the speed threshold, the driver's eyelid closure behavior will be re-evaluated and an alarm will be triggered every preset time interval.

8. An eyelid closure detection device, characterized in that, include: The perception information acquisition module is used to perform perception detection on people in a predetermined number of frames of historical images and to mark the positions of binocular reference points in the predetermined number of frames of historical images, wherein the binocular reference points include a left eye reference point and a right eye reference point. The normalization calculation module is used to calculate the distance between the upper and lower eyelids of each eye in the predetermined number of historical images based on the position of the binocular reference points, to obtain the eye width in each frame of the historical image, and to filter the eye widths in the predetermined number of historical images to obtain the maximum eye width, and to normalize the maximum eye width to obtain a normalization interval, wherein the normalization interval is obtained by using the minimum value of the eye width and the maximum eye width; The width normalization module is used to calculate the distance between the positions of the binocular reference points detected by the perception model for the person in each frame of the current image, to obtain the current eye width, and to map the current eye width to the normalization interval to obtain the current width normalization value. The maximum eye width is mapped to the normalized maximum value, and the eye width when the eyes are closed is mapped to the normalized minimum value, so that the current eye width is magnified when mapped to the normalization interval. The width calculation module is used to sort multiple current width normalized values ​​within a preset time period to obtain a width time sequence queue, and to perform weighted calculation on the current width normalized values ​​in the width time sequence queue to obtain the average width of the width time sequence queue. The behavior determination module is used to compare the average width with a width threshold. When the average width is less than the width threshold, the module determines that the person's eye state is closed.

9. A vehicle, characterized in that, The vehicle includes the eyelid closure detection device as described in claim 8.

10. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed, the computer performs the eyelid closure detection method as described in any one of claims 1-5 or the driver eyelid closure alarm method as described in any one of claims 6-7.

11. A computer device comprising a processor and a memory, the memory storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the eyelid closure detection method as described in any one of claims 1-5 or the driver eyelid closure alarm method as described in any one of claims 6-7.

Images