KR20240011544A - Method and apparatus for determining whether seat belt is worn - Google Patents

Abstract

acquiring an image including a seated person; determining a first bounding box corresponding to the upper body area of the seated person; detecting a seat belt and determining a second bounding box containing the detected seat belt; and determining whether the seated person is wearing a seat belt based on a degree of overlap between the first bounding box and the second bounding box. A method and device for determining whether a seat belt is worn is introduced.

Description

Method and device for determining whether a seat belt is worn {METHOD AND APPARATUS FOR DETERMINING WHETHER SEAT BELT IS WORN}

The present invention relates to a method and device for determining whether a seated person is wearing a seat belt based on analysis of image information.

Most vehicles today are equipped with seat belts to ensure the safety of occupants in the event of a vehicle crash. These seat belts, also known as seat belts, minimize injuries to occupants from collision accidents by preventing occupants from being suddenly separated from the seat and colliding with the vehicle's internal structure or being thrown out of the vehicle due to the impact that occurs during a vehicle collision. perform its role.

These seat belts restrain the occupant's body to prevent the occupant from leaving the seat, which causes pressure to be transmitted and the occupant feels uncomfortable wearing the seat belt. Accordingly, even though seat belts are provided, a problem has arisen where passengers do not wear seat belts and suffer serious injuries.

To solve this problem, conventional technologies that encourage passengers to wear seat belts have been developed. One example of the prior art is a technology that induces seat belt wearing using weight information. This technology determines the presence or absence of an occupant by detecting the load applied to the seat through a switch sensor attached to the seating surface at the top of the seat, and determines whether the seat belt is fastened through a buckle sensor attached to the sheath belt. Through this, it is possible to recognize cases where the seat belt is not fastened even though there is an occupant, and it is possible to encourage people to wear the seat belt through notifications, etc.

However, in order to avoid wearing a seat belt, this technology uses a separate tool such as a fake buckle to make the sensor recognize that the seat belt is fastened, or the seat belt is worn inappropriately, contrary to its original usage, so that it does not restrain the body. Even in this case, it is recognized that the seat belt is fastened. Therefore, there is a problem that the function for encouraging the seat belt to be worn does not work because the occupant is judged to be wearing a seat belt even though he or she is not actually wearing the seat belt.

KR 10-0844427 B1

The present invention relates to a method and device for determining whether a seated person is wearing a seat belt, which allows more accurate determination of whether a seated person is wearing a seat belt based on analysis of image information.

The technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below. You will be able to.

In order to solve the above technical problem, a method of determining whether a seat belt is worn according to an embodiment of the present invention includes the steps of acquiring an image including a seated person; determining a first bounding box corresponding to the upper body area of the seated person from the acquired image; detecting a seat belt from the acquired image and determining a second bounding box containing the detected seat belt; and determining whether the seated person is wearing a seat belt based on a degree of overlap between the first bounding box and the second bounding box.

For example, determining the first bounding box may include detecting at least one preset point included in the upper body area of the seated person from the image; and determining the first bounding box based on the detected point.

For example, the preset point may include points corresponding to both shoulders and the center of the abdomen of the seated person's upper body.

For example, determining the first bounding box may include determining a width of the first bounding box based on points corresponding to each of the two shoulders; and determining the height of the first bounding box based on at least one of the points corresponding to each of the two shoulders and a point corresponding to the center of the abdomen.

For example, if some of the preset points are not detected, deriving points that were not detected based on other detected points, the step of determining the first bounding box may include deriving points that were not detected. The first bounding box can be determined by additionally considering .

For example, determining the second bounding box may include extracting seat belt characteristic information from the acquired image; Obtaining classification information from the extracted seat belt characteristic information; and detecting a seat belt based on the obtained classification information.

For example, the step of determining whether the seat belt is worn may include comparing the degree of overlap with a preset threshold to determine whether the seat belt is worn.

For example, the step of determining whether the seat belt is worn may include determining that the seated person is not wearing a seat belt when the overlap degree is less than or equal to a preset threshold.

For example, determining the first bounding box, determining the second bounding box, or a combination thereof may be performed by inputting the acquired image into a predetermined artificial intelligence learning model.

For example, the artificial intelligence learning model is an artificial intelligence algorithm and may include at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof.

In order to solve the above technical problem, an apparatus for determining whether a seat belt is worn according to an embodiment of the present invention includes a camera for photographing the inside of a vehicle; and determining a first bounding box and a second bounding box from the image including the seated person obtained through the camera, and whether the seated person is wearing a seat belt based on the degree of overlap between the first bounding box and the second bounding box. and a controller that determines, wherein the first bounding box corresponds to an upper body area of the seated person, and the second bounding box includes a seat belt detected from the acquired image through the controller.

For example, the controller may detect at least one preset point included in the upper body area of the seated person from the image and determine the first bounding box based on the detected point.

For example, the preset point may include points corresponding to both shoulders and the center of the abdomen of the seated person's upper body.

For example, the controller determines the width of the first bounding box based on the points corresponding to each of the two shoulders, and determines the width of the first bounding box to at least one of the points corresponding to each of the two shoulders and a point corresponding to the center of the abdomen. Based on this, the height of the first bounding box can be determined.

For example, if some of the preset points are not detected, the controller may derive a point that was not detected based on other detected points and determine the first bounding box by additionally considering the derived points. .

For example, the controller may extract seat belt characteristic information from the acquired image, obtain classification information from the extracted seat belt specific information, and detect the seat belt based on the obtained classification information. there is.

For example, the controller may determine whether a seat belt is worn by comparing the degree of overlap with a preset threshold.

For example, the controller may determine that the seated person is not wearing a seat belt when the degree of overlap is less than or equal to a preset threshold.

For example, the controller may determine at least one of the first bounding box, the second bounding box, or a combination thereof by inputting the acquired image into a predetermined artificial intelligence learning model.

For example, the artificial intelligence learning model is an artificial intelligence algorithm and may include at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof.

The object of the present invention is not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

The method for determining whether a seat belt is worn or not according to at least one embodiment of the present invention configured as described above more accurately determines whether a seat belt is worn based on analysis of image information, and determines whether the seat belt is properly worn according to its original usage. You can determine whether it has been done.

Furthermore, by determining whether or not a seat belt is worn using the bounding box corresponding to the seated person and the seat belt on the image information, the accuracy of image information analysis can be improved and the judgment speed can be further improved.

The effects that can be obtained from the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below. will be.

1 is a diagram showing the configuration of a device for determining whether a seat belt is worn according to an embodiment of the present invention.
Figure 2 is a diagram illustrating a process for deriving the degree of overlap between bounding boxes to determine whether a seat belt is worn, according to an embodiment of the present invention.
Figure 3 is a diagram for explaining a method of determining whether a seat belt is worn based on the degree of overlap, according to an embodiment of the present invention.
Figure 4 is a flowchart of a method for determining whether a seat belt is worn according to an embodiment of the present invention.

Specific structural and functional descriptions of the embodiments of the present invention disclosed in the present specification or application are merely illustrative for the purpose of explaining the embodiments according to the present invention, and the embodiments according to the present invention are implemented in various forms. It may be possible and should not be construed as being limited to the embodiments described in this specification or application.

Since the embodiments according to the present invention can make various changes and have various forms, specific embodiments will be illustrated in the drawings and described in detail in the specification or application. However, this is not intended to limit the embodiments according to the concept of the present invention to a specific disclosed form, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the technical field to which the present invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and unless clearly defined in this specification, should not be interpreted as having an ideal or excessively formal meaning. .

Hereinafter, embodiments disclosed in the present specification will be described in detail with reference to the attached drawings. However, identical or similar components will be assigned the same reference numbers regardless of reference numerals, and duplicate descriptions thereof will be omitted.

The suffixes “module” and “part” for components used in the following description are given or used interchangeably only for the ease of preparing the specification, and do not have distinct meanings or roles in themselves.

In describing the embodiments disclosed in this specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in this specification, the detailed descriptions will be omitted. In addition, the attached drawings are only for easy understanding of the embodiments disclosed in this specification, and the technical idea disclosed in this specification is not limited by the attached drawings, and all changes included in the spirit and technical scope of the present invention are not limited. , should be understood to include equivalents or substitutes.

Terms containing ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to the other component, but that other components may exist in between. It should be. On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this specification, terms such as “comprise” or “have” are intended to indicate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

A controller is a communication device that communicates with other controllers or sensors to control the function it is in charge of, a memory that stores the operating system, logic commands, input/output information, etc., and a device that performs the judgments, calculations, and decisions necessary to control the function it is in charge of. It may include more than one processor.

In embodiments of the present invention, when determining whether or not a seat belt is worn based on analysis of image information, whether or not a seat belt is worn is determined based on the bounding box corresponding to the seated person and the seat belt, thereby determining whether or not the seat belt is worn. We propose a method and device for determining whether a seat belt is worn, which accurately determines whether the seat belt is worn properly according to its intended use.

Before explaining the method for determining whether a seat belt is worn according to the present invention, a device for determining whether a seat belt is worn according to an embodiment of the present invention will first be described with reference to FIG. 1 .

Figure 1 is a diagram showing the configuration of a device for determining whether a seat belt is worn according to an embodiment of the present invention.

Referring to FIG. 1, an apparatus 100 for determining whether a seat belt 110 is worn according to an embodiment includes a camera 140 and a controller 150. Figure 1 mainly shows components related to embodiments of the present invention, and it is obvious to those skilled in the art that more components may be included in the actual implementation of the decision device 100. Below, each component will be described.

First, the camera 140 performs the function of photographing the interior of the vehicle and transmitting the photographed image to the controller 150, etc. For the convenience of determining whether or not the seat belt 110 is worn, it is appropriate that the camera 140 be installed at the front of the vehicle and shoot toward the rear so that the front of the seated person can be clearly recognized. In this case, the camera 140 that photographs the interior of the vehicle may be implemented as an in-cabin camera. Meanwhile, the in-cabin camera is mounted inside the vehicle and can check the status of the seated person in real time. If the camera 140 is implemented as an in-cabin camera, it continuously checks whether the seat belt 110 is worn by using the image provided in real time. Monitoring becomes possible.

Meanwhile, the controller 150 obtains an image including the seated person from the camera 140 and determines whether the seated person is wearing the seat belt 110 through the image. The controller 150 determines the first bounding box 210 and the second bounding box 220 from the acquired image and determines whether the seated person is wearing the seat belt 110 based on the degree of overlap between the determined bounding boxes. Here, the first bounding box 210 corresponds to the upper body area of the seated person on the acquired image, and the second bounding box 220 includes the seat belt 110 detected from the acquired image. A more detailed description of the determination by the controller 150 of whether the seated person is wearing the seat belt 110 will be described below with reference to FIGS. 2 and 3 .

FIG. 2 is a diagram illustrating a process for deriving a degree of overlap between bounding boxes to determine whether a seat belt is worn, according to an embodiment of the present invention.

Referring to FIG. 2 , a seated person, a seat belt 110, a first bounding box 210, a second bounding box 220, and an overlapping area 230 are displayed on the image acquired from the camera 140.

First, the process of determining the first bounding box 210 corresponding to the upper body area of the seated person from the acquired image will be described. The controller 150 may detect at least one preset point included in the upper body area of the seated person from the image acquired to determine the first bounding box 210. In this case, each point may correspond to each point of the upper body of the seated person shown on the image. In particular, in order to ensure that the first bounding box 210 corresponds to the upper body area of the seated person, the left shoulder 211, which is the boundary of the upper body, It can be set to correspond to the right shoulder 212 and the center of the abdomen 213.

The controller 150 may determine the first bounding box 210 based on each detected point. In particular, when points corresponding to both shoulders 211, 212 and the center of the abdomen of the seated person's upper body shown in the acquired image are detected, the points corresponding to the left shoulder 211 and the points corresponding to the right shoulder 212 are used. 1 The width of the bounding box 210 can be determined. Additionally, the height of the first bounding box 210 may be determined based on at least one of the points corresponding to both shoulders 211 and 212 and the point 213 corresponding to the center of the abdomen. For example, the horizontal distance between the points corresponding to both shoulders 211 and 212 is the width of the first bounding box 210, and the distance between the points corresponding to both shoulders 211 and 212 and the abdominal central controller 150 is The vertical distance may be determined by the height of the first bounding box 210. By determining the first bounding box 210 in this way, it is possible to prevent the first bounding box 210 from being determined to be too small or large compared to the upper body area of the seated person in the image so that it can appropriately correspond to the upper body area of the seated person. There will be.

Meanwhile, if part of the seated person's upper body is obscured by the seated person's posture or other obstacles, some of the preset points may not be detected from the acquired image. In this case, the controller 150 may derive points that were not detected based on the remaining detected points, and determine the first bounding box 210 by considering the derived points together with the detected points. For example, if the left shoulder 211 of the seated person is obscured and the corresponding point is not detected, the horizontal distance between the point corresponding to the center of the abdomen 213 or the solar plexus and the right shoulder 212 is calculated, and both shoulders Under the assumption that is symmetrical, a point located away from the point corresponding to the center of the abdomen 213 or the solar plexus by the above-mentioned horizontal distance to the left can be derived as the point corresponding to the left shoulder 211. Additionally, undetected points can also be derived by utilizing separate human body information that has been stored or received differently.

Meanwhile, the controller 150 may detect the seat belt 110 from the acquired image and determine the second bounding box 220 including the detected seat belt 110. To this end, the controller 150 can extract characteristic information of the seat belt 110 from the acquired image, obtain classification information from the extracted specific information, and detect the seat belt based on the obtained classification information. You can. Here, the seat belt 110 characteristic information may include the color, shape, and location of the seat belt 110, and the classification information obtained through this allows the seat belt 110 and other objects to be distinguished in the image. .

Meanwhile, determination of the first bounding box 210 and the second bounding box 210 can be performed by inputting the acquired image into a predetermined artificial intelligence learning model. Here, the artificial intelligence learning model is an artificial intelligence algorithm and may include at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof.

In particular, Human Pose Estimation of an artificial intelligence learning model can be used to determine the first bounding box 210. Human Pose Estimation detects a plurality of key points corresponding to important human body parts and estimates the target human pose through them. Among the detected key points, those included in the upper body of the seated person are used in an embodiment of the present invention. It can be used as points for determining the first bounding box 210. In addition, in detecting the seat belt 110, classification of an artificial intelligence learning model is used to improve detection accuracy, and the first bounding box 210 and the second bounding box 220 are corrected to have more accurate values. To do this, Bounding Box Regression of the artificial intelligence learning model can be used.

When the first bounding box 210 and the second bounding box 220 are determined through this process, the controller 150 determines their degree of overlap. Here, the degree of overlap may be calculated based on the area of the area 230 included in both the first bounding box 210 and the second bounding box 220 in the acquired image.

As described with reference to FIG. 2, by determining whether the seat belt 110 is worn through analysis of the image acquired through the camera 140, the seat belt 110 is recognized as fastened by a buckle sensor, etc., but the seated person is actually Even when the seat belt 110 is not worn, it can be determined that the seat belt 110 is not worn. Furthermore, through this, it is possible to encourage actual wearing of the seat belt 110.

On the other hand, when determining whether or not the seat belt 110 is worn through segmentation of the seat belt 110 area on the acquired image, a huge amount of information equal to the number of pixels in the image is required, which may overload the controller 150. You can. On the other hand, by using the first bounding box 210 and the second bounding box 220 according to embodiments of the present invention, whether the seat belt 110 is worn only with information about the four vertices, width, and height of the bounding box. It becomes possible to judge. This reduces the amount of information required for judgment and improves judgment speed.

Hereinafter, the process of determining whether the seat belt 110 is worn based on the degree of overlap calculated through the above process will be described with reference to FIG. 3.

Figure 3 is a diagram for explaining a method of determining whether a seat belt is worn based on the degree of overlap, according to an embodiment of the present invention.

Referring to FIG. 3, in one embodiment, in order to explain a method of determining whether a seat belt is worn, a graph may be displayed with the degree of overlap as the vertical axis and time as the horizontal axis. The displayed graph shows an approximate trend to aid understanding and does not represent exact values.

first. On the graph, you can see that the degree of overlap does not have a fixed value but changes. Although the degree of overlap is not determined by time, since the seated person is not a fixed object, the area of the seated person and the seat belt 110 appearing on the image changes according to the movement of the seated person, and thus the first bounding box This is because the degree of overlap between 210 and the second bounding box 220 changes.

In one embodiment of the present invention, to determine whether or not the seat belt 110 is worn, the camera 140 continuously photographs the seated person and transmits the images to the controller 150 in real time, and the controller 150 receives the images to determine whether the seat belt 110 is worn. It can be determined whether the person is wearing the seat belt 110. Through this, it is possible to continuously monitor whether the seated person is wearing the seat belt 110, which can change as shown in the graph.

Meanwhile, the controller 150 compares the degree of overlap between the first bounding box 210 and the second bounding box 220 with a preset threshold, and determines whether the seat belt 110 is worn based on the comparison result. can do.

In this case, the first bounding box corresponds to the upper body area of the seated person, and the second bounding box includes the seat belt 110, so the higher the overlap, the more likely it is that the seated person is wearing the seat belt 110. This becomes higher. Accordingly, the controller 150 determines that the seated person is wearing the seat belt 110 when the degree of overlap exceeds the preset threshold, and determines that the seated person is not wearing the seat belt 110 when the degree of overlap is less than the preset threshold. It can be judged that

The contents described above are expressed in a flowchart as shown in FIG. 4. Hereinafter, a method for determining whether the seat belt 110 is worn according to an embodiment of the present invention will be described with reference to FIG. 4.

Figure 4 is a flowchart of a method for determining whether a seat belt is worn according to an embodiment of the present invention.

Referring to FIG. 4, first, the camera 140 captures an image of the inside of the vehicle including the seated person, and the controller 150 acquires the image (S401). In this case, for the convenience of determining whether or not the seat belt 110 is worn, it is appropriate for the camera 140 to be installed at the front of the vehicle and shoot toward the rear so that the front of the seated person can be clearly recognized. Meanwhile, in this case, the camera 140 that photographs the interior of the vehicle may be implemented as an in-cabin camera. The in-cabin camera is mounted inside the vehicle and can check the status of the seated person in real time. If the camera 140 is implemented as an in-cabin camera, it is possible to continuously monitor whether the seat belt 110 is worn using images provided in real time. It becomes possible.

The controller 150 that acquired the image detects at least one preset point included in the upper body area of the seated person from the acquired image to determine the first bounding box 210 (S402). In this case, each point may correspond to each point of the upper body of the seated person shown on the image. In particular, in order to ensure that the first bounding box 210 corresponds to the upper body area of the seated person, the left shoulder 211, which is the boundary of the upper body, It can be set to correspond to the right shoulder 212 and the center of the abdomen 213.

Thereafter, the controller 150 determines the first bounding box 210 based on the detected point (S404). In this case, when points corresponding to both shoulders 211, 212 and the center of the abdomen of the seated person's upper body shown in the acquired image are detected, based on the points corresponding to the left shoulder 211 and the points corresponding to the right shoulder 212 Determine the width of the first bounding box 210, and determine the height of the first bounding box 210 based on at least one of the points 211 and 212 corresponding to both shoulders and the point 213 corresponding to the center of the abdomen. You can. For example, the horizontal distance between the points corresponding to both shoulders 211 and 212 is the width of the first bounding box 210, and the distance between the points corresponding to both shoulders 211 and 212 and the abdominal central controller 150 is The vertical distance may be determined by the height of the first bounding box 210. By determining the first bounding box 210 in this way, the controller 150 prevents the first bounding box 210 from being determined to be too small or large compared to the upper body area of the seated person in the image and appropriately corresponds to the upper body area of the seated person. You can make it happen.

Meanwhile, if part of the seated person's upper body is obscured by the seated person's posture or other obstacles, some of the preset points may not be detected from the acquired image. In this case, the controller 150 may derive the undetected point based on other detected points, and determine the first bounding box 210 by considering the derived point together with the detected point. For example, if the left shoulder 211 of the seated person is obscured and the corresponding point is not detected, the horizontal distance between the point corresponding to the center of the abdomen 213 or the solar plexus and the right shoulder 212 is calculated, and both shoulders Under the assumption that is symmetrical, a point located away from the point corresponding to the center of the abdomen 213 or the solar plexus by the above-mentioned horizontal distance to the left can be derived as the point corresponding to the left shoulder 211. Additionally, undetected points can also be derived by utilizing separate human body information that has been stored or received differently.

Additionally, the controller 150 detects the seat belt 110 from the acquired image to determine the second bounding box 220 (S403). The controller 150 may extract characteristic information of the seat belt 110 from the acquired image, obtain classification information from the extracted specific information, and detect the seat belt 110 based on the obtained classification information. . Here, the seat belt 110 characteristic information may include the color, shape, and location of the seat belt 110, and the classification information obtained through this allows the seat belt 110 and other objects to be distinguished in the image. .

When the seat belt 110 is detected, the controller 150 determines a second bounding box including the detected seat belt 110 (S405). In this case, the second bounding box includes the seat belt 110. It is appropriate to determine the minimum area as a rectangular area.

Meanwhile, the above processes (S402 to S405) can be performed by inputting the acquired images into a predetermined artificial intelligence learning model. Here, the artificial intelligence learning model is an artificial intelligence algorithm and may include at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof. In particular, Human Pose Estimation of an artificial intelligence learning model can be used to determine the first bounding box 210. Human Pose Estimation detects a plurality of key points corresponding to important human body parts and estimates the target human pose through them. Among the detected key points, those included in the upper body of the seated person are used in an embodiment of the present invention. It can be used as points for determining the first bounding box 210. In addition, in detecting the seat belt 110, classification of an artificial intelligence learning model is used to improve detection accuracy, and the first bounding box 210 and the second bounding box 220 are corrected to have more accurate values. To do this, Bounding Box Regression of the artificial intelligence learning model can be used.

When the first bounding box 210 and the second bounding box 220 are determined, the controller 150 determines the degree of overlap between the two boxes 210 and 220 (S406), and connects the seat belt 110 of the seated person based on the degree of overlap. You can decide whether to wear it or not. Here, the degree of overlap may be calculated based on the area of the area 230 included in both the first bounding box 210 and the second bounding box 220 in the acquired image.

The controller 150 can determine whether the seat belt 110 is worn by comparing the degree of overlap with a preset threshold. If the degree of overlap exceeds the preset threshold (Yes in S407), the seat belt is worn by the seat belt. It can be determined that the seat belt 110 is worn (S409), and if the degree of overlap is less than a preset threshold (No in S407), it can be determined that the seated person is not wearing the seat belt 110 (S408).

The method for determining whether a seat belt is worn or not according to at least one embodiment of the present invention configured as described above more accurately determines whether a seat belt is worn based on analysis of image information, and determines whether the seat belt is properly worn according to its original usage. You can determine whether it has been done.

Furthermore, by determining whether or not a seat belt is worn using the bounding box corresponding to the seated person and the seat belt on the image information, the accuracy of image information analysis can be improved and the judgment speed can be further improved.

Although the present invention has been shown and described in relation to specific embodiments of the present invention as described above, it is understood that the present invention can be variously improved and changed without departing from the technical spirit of the present invention provided by the following claims. This will be self-evident to those skilled in the art.

140: camera
150: controller
210: first bounding box
220: second bounding box

Claims (20)

acquiring an image including a seated person;
determining a first bounding box corresponding to the upper body area of the seated person from the acquired image;
detecting a seat belt from the acquired image and determining a second bounding box containing the detected seat belt; and
Determining whether the seated person is wearing a seat belt based on a degree of overlap between the first bounding box and the second bounding box. In claim 1,
The step of determining the first bounding box is,
detecting at least one preset point included in the upper body area of the seated person from the image; and
A method for determining whether a seat belt is worn, comprising: determining the first bounding box based on the detected point. In claim 2,
The preset points are,
A method for determining whether a seat belt is worn, comprising points corresponding to both shoulders and the center of the abdomen of the seated person's upper body. In claim 3,
The step of determining the first bounding box is,
determining a width of the first bounding box based on points corresponding to each of the two shoulders; and
Determining the height of the first bounding box based on at least one of the points corresponding to both shoulders and a point corresponding to the center of the abdomen. In claim 2,
If some of the preset points are not detected, deriving points that were not detected based on the remaining detected points; further comprising:
The step of determining the first bounding box is,
A method for determining whether a seat belt is worn, characterized in that the first bounding box is determined by additionally considering the derived points. In claim 1,
The step of determining the second bounding box is,
extracting seat belt characteristic information from the acquired image;
Obtaining classification information from the extracted seat belt characteristic information; and
A method for determining whether a seat belt is worn, comprising: detecting a seat belt based on the obtained classification information. In claim 1,
The step of determining whether the seat belt is worn is,
A method for determining whether a seat belt is worn by comparing the degree of overlap with a preset threshold to determine whether a seat belt is worn. In claim 7,
The step of determining whether the seat belt is worn is,
A method for determining whether a seat belt is worn, comprising: determining that the seated person is not wearing a seat belt when the degree of overlap is less than or equal to a preset threshold. In claim 1,
Determining the first bounding box, determining the second bounding box, or a combination thereof,
A method of determining whether a seat belt is worn, characterized in that it is performed by inputting the acquired image into a predetermined artificial intelligence learning model. In claim 9,
The artificial intelligence learning model is an artificial intelligence algorithm, and includes at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof. Cameras that film the interior of a vehicle; and
A first bounding box and a second bounding box are determined from the image including the seated person obtained through the camera, and whether the seated person is wearing a seat belt is determined based on the degree of overlap between the first bounding box and the second bounding box. Includes a controller that determines,
The first bounding box is,
Corresponds to the upper body area of the seated person,
The second bounding box is,
A device for determining whether a seat belt is worn, comprising a seat belt detected from the acquired image through the controller. In claim 11,
The controller is,
A device for determining whether a seat belt is worn, characterized in that detecting at least one preset point included in the upper body area of the seated person from the image and determining the first bounding box based on the detected point. In claim 12,
The preset points are,
A device for determining whether a seat belt is worn, comprising points corresponding to both shoulders and the center of the abdomen of the seated person's upper body. In claim 13,
The controller is,
Determine the width of the first bounding box based on the points corresponding to each of the two shoulders,
A device for determining whether a seat belt is worn, characterized in that the height of the first bounding box is determined based on at least one of the points corresponding to each of the two shoulders and a point corresponding to the center of the abdomen. In claim 12,
The controller is,
When some of the preset points are not detected, a point that was not detected is derived based on other detected points, and the first bounding box is determined by additionally considering the derived points. Device. In claim 11,
The controller is,
Determining whether or not a seat belt is worn by extracting seat belt characteristic information from the acquired image, obtaining classification information from the extracted seat belt specific information, and detecting the seat belt based on the obtained classification information. Device. In claim 11,
The controller is,
A device for determining whether a seat belt is worn by comparing the degree of overlap with a preset threshold to determine whether the seat belt is worn. In claim 17,
The controller is,
A device for determining whether a seat belt is worn, characterized in that it is determined that the seated person is not wearing a seat belt when the degree of overlap is less than a preset threshold. In claim 11,
The controller is,
A device for determining whether a seat belt is worn, characterized in that it determines at least one of the first bounding box, the second bounding box, or a combination thereof by inputting the acquired image into a predetermined artificial intelligence learning model. In claim 19,
The artificial intelligence learning model is an artificial intelligence algorithm, and includes at least one of machine learning, neural network, deep learning, classification algorithm, or a combination thereof. A device for determining whether a seat belt is worn.

Images