Method and system for generating object control data using an ai-based object detection model

Abstract

The present invention relates to a method and system for automatically generating object control data using an artificial intelligence-based object detection model. The method for automatically generating object control data according to the present invention comprises the steps of: collecting sensor data of a control target area using a sensor; generating object detection data using an artificial intelligence-based object detection model based on the sensor data and storing it in a database; generating interest area setting data based on the object detection data; and generating an interest area based on the object detection data and the interest area setting data.

Description

Technology Field

[0001] The present invention relates to a method and system for automatically generating object control data using an artificial intelligence-based object detection model. Background Technology

[0002] Research is being conducted on methods and devices for generating information regarding objects subject to control using artificial intelligence capable of detecting objects. For example, a parking control system is being developed that incorporates a function to detect the presence of vehicles within a parking lot by applying artificial intelligence to sensor data collected using radar and cameras. In other words, when a vehicle, which is the object subject to control, occupies a specific area (e.g., a parking space), the parking control system uses sensor data and artificial intelligence to detect and indicate that the specific area is occupied.

[0003] However, in such monitoring systems, the detection range is often manually set by humans. In other words, to enable artificial intelligence to automatically detect whether an object occupies a specific area of ​​the monitored space, humans frequently manually define the detection range from sensor data, such as images; however, this process leads to inefficiency and prolongs the time required to normalize the monitoring system. Prior art literature

[0004] (Patent Document 0001) KR 10-2310567 B1 (2021.10.1.)(Patent Document 0002) KR 10-2024-0145342 A (2024.10.7.) The problem to be solved

[0005] The present invention aims to provide a method and system for automatically generating object control data, which automatically sets a region of interest in sensor data based on artificial intelligence learned through an unsupervised learning algorithm and automatically generates control data using an object detection model.

[0006] The objectives of the present invention are not limited to those mentioned above, and other unmentioned objectives will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0007] A method for automatically generating object control data according to an embodiment of the present invention comprises: a step in which an object control data automatic generation system collects sensor data of a control target area using a sensor; a step in which the system generates object detection data using an artificial intelligence-based object detection model based on the sensor data and stores it in a database; a step in which the system generates interest area setting data based on the object detection data; and a step in which the system generates an interest area by removing a portion of the interest area setting data based on the object detection data.

[0008] In one embodiment of the present invention, the method for automatically generating object control data further includes the step of the system generating object control data for the control target area based on object detection data corresponding to recent sensor data among the sensor data and the region of interest.

[0009] In one embodiment of the present invention, the sensor is one of a camera, radar, and lidar.

[0010] In one embodiment of the present invention, the object detection data is data represented by two intersecting line segments, and the location of the detected object is the intersection point of the two line segments.

[0011] In one embodiment of the present invention, the region of interest setting data is data represented by a baseline.

[0012] In one embodiment of the present invention, the control target area is a parking lot, the area of ​​interest is a parking space, and the object control data is data indicating whether a vehicle is parked in the parking space.

[0013] In one embodiment of the present invention, the step of generating the interest region setting data is,

[0014] The above system may include applying a clustering model to the object detection data collected over a predetermined period to classify the object detection data into a plurality of groups, and generating the region of interest setting data based on the object detection data included in each group.

[0015] An object control data automatic generation system according to one embodiment of the present invention includes: a memory for storing computer-readable commands; and at least one processor implemented to execute said commands.

[0016] The above-mentioned at least one processor is configured to collect sensor data of a controlled area by executing the above commands, generate object detection data using an artificial intelligence-based object detection model based on the sensor data and store it in a database, generate region of interest setting data based on the object detection data, and generate a region of interest by removing a portion of the region of interest setting data based on the object detection data.

[0017] In one embodiment of the present invention, the at least one processor is configured to generate object control data for the control target area based on object detection data corresponding to recent sensor data among the sensor data and the region of interest.

[0018] In one embodiment of the present invention, the sensor is one of a camera, radar, and lidar.

[0019] In one embodiment of the present invention, the object detection data is data represented by two intersecting line segments, and the location of the detected object is the intersection point of the two line segments.

[0020] In one embodiment of the present invention, the region of interest setting data is data represented by a baseline.

[0021] In one embodiment of the present invention, the control target area is a parking lot, the area of ​​interest is a parking space, and the object control data is data indicating whether a vehicle is parked in the parking space.

[0022] In one embodiment of the present invention, the at least one processor is,

[0023] It can be configured to apply a clustering model to the object detection data collected over a predetermined period to classify the object detection data into multiple groups, and to generate the region of interest setting data based on the object detection data included in each group. Effects of the invention

[0024] According to one embodiment of the present invention, the inefficiency of having to manually set the area of ​​interest of sensor data during the construction of an artificial intelligence-based control system can be eliminated.

[0025] In addition, according to one embodiment of the present invention, the control data generation process can be advanced and lightweight.

[0026] The effects obtainable from the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0027] FIGS. 1 and FIGS. 2 are example drawings illustrating the process of determining whether an object occupies a region of interest. FIG. 3 is a block diagram showing the configuration of an object control data automatic generation system according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method for automatically generating object control data according to an embodiment of the present invention. FIGS. 5a to 5d are example drawings of sensor data collected through an IP camera. FIGS. 6a to 6d are example drawings of accumulated object detection data. FIGS. 7 and 8 are drawings showing a baseline for object control and accumulated object detection data displayed on sensor data. FIGS. 9a through 9d are example drawings of the region of interest. FIGS. 10a and FIGS. 10b are example drawings showing the diagonal components of detection data accumulated in sensor data collected through an IP camera. FIGS. 11a and 11b are drawings showing the line segment connecting the center of an adjacent region of interest and the diagonal of the object detection data in the sensor data collected through an IP camera. FIG. 12 is a diagram showing a line segment connecting the centers of adjacent regions of interest. FIG. 13 is a diagram showing the angle formed by the extension of the line segment connecting the centers of adjacent regions of interest and the diagonal of the object detection data. FIG. 14 is a diagram illustrating a case where the diagonal of object detection data serving as a control standard is set differently depending on the region of interest. FIG. 15 is an example diagram of a region of interest and control data generated by the method for automatically generating object control data according to the present invention. FIGS. 16a and FIGS. 16b are examples of real-time object control screens. Specific details for implementing the invention

[0028] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. Meanwhile, the terms used in this specification are for describing the embodiments and are not intended to limit the present invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms "comprises" and / or "comprising" as used in this specification do not exclude the presence or addition of one or more other components, steps, actions, and / or elements in addition to the mentioned components, steps, actions, and / or elements.

[0029] Terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. These terms may be used for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component.

[0030] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. Conversely, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between. Other expressions describing the relationship between components, such as "between" and "exactly between," or "adjacent to" and "directly adjacent to," should be interpreted in the same way.

[0031] In describing the present invention, detailed descriptions of related prior art are omitted if it is determined that such descriptions may unnecessarily obscure the essence of the invention.

[0032] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In order to facilitate an overall understanding in describing the present invention, the same reference numerals will be used for the same means regardless of the drawing number.

[0033] Figures 1 and 2 are example drawings illustrating the process of determining whether an object occupies a region of interest.

[0034] An object control data automatic generation system (100) according to one embodiment of the present invention can determine whether an object (e.g., a vehicle) occupies a Region of Interest (ROI) in a form such as the example of FIG. 1 or FIG. 2.

[0035] FIG. 1 is an example drawing in which an area of ​​interest and an object are marked with rectangles on an image of a control target area (parking lot). In FIG. 1, the area of ​​interest (parking space) is marked with a yellow rectangle or a green rectangle, and the object (vehicle) is marked with a red rectangle. The object control data automatic generation system (100) can determine that the parking space is occupied if there is an overlapping area between the area of ​​interest (parking space) and the area of ​​the object (vehicle).

[0036] Figure 2 is an example drawing in which, on an image of a control target area (parking lot), the area of ​​interest (parking space) is indicated by a yellow or red line, and the detected object is indicated by two intersecting line segments. Here, the intersection point of the two line segments indicates the location of the object. The object control data automatic generation system (100) can determine whether the parking space is occupied based on whether an object exists in the area of ​​interest (parking space).

[0037] However, when a person manually sets the area of ​​interest in FIG. 1 or FIG. 2, there is a problem of inefficiency occurring in the process of generating control data, including whether an object is occupied. The present invention applies artificial intelligence technology to automate, enhance, and lighten the task of setting the area of ​​interest.

[0038] FIG. 3 is a block diagram showing the configuration of an object control data automatic generation system according to one embodiment of the present invention.

[0039] The object control data automatic generation system (100) may include at least one processor (110), a communication device (120), a memory (130), a storage device (140), and a sensor (150) that communicate via a bus (160). The object control data automatic generation system (100) illustrated in FIG. 3 is according to one embodiment, and the components of the object control data automatic generation system (100) according to the present invention are not limited to the embodiment illustrated in FIG. 3 and may be added, changed, or deleted as needed. For example, the object control data automatic generation system (100) may further include an output interface device such as a display.

[0040] The processor (110) may be a central processing unit (CPU) or a semiconductor device that executes computer-readable instructions stored in memory (130) or storage device (140). Memory (130) and storage device (140) may include various forms of volatile or non-volatile storage media. For example, memory (130) may include read-only memory (ROM) and random access memory (RAM). In the embodiments of this description, memory (130) may be located inside or outside the processor (110), and memory (130) may be connected to the processor (110) through various known means. Memory (130) is various forms of volatile or non-volatile storage media, and for example, memory (130) may include read-only memory (ROM) or random access memory (RAM).

[0041] Accordingly, embodiments of the present invention may be implemented as a method implemented on a computer or as a non-transient computer-readable medium in which computer-executable instructions are stored. In one embodiment, when executed by a processor (110), the computer-readable instructions may perform the method for automatically generating object control data according to the present invention.

[0042] The communication device (120) can transmit or receive wireless signals. The communication device (120) can transmit and receive data via wireless communication with a communication device mounted on an external computing device.

[0043] In addition, the storage device (140) stores various data through a database.

[0044] The sensor (150) may include one or a combination of a camera, radar, and lidar. Sensor data collected through the sensor (150) may be an image and / or point cloud. The sensor (150) continuously collects sensor data according to a fixed period, and the processor (110) may store the sensor data in memory (130) or a storage device (140).

[0045] In addition, the method for automatically generating object control data according to an embodiment of the present invention may be implemented in the form of program instructions that can be executed through various computer means and may be recorded on a computer-readable medium.

[0046] The above computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the computer-readable medium may be specially designed and configured for embodiments of the present invention, or they may be known and available to a person skilled in the art of computer software. The computer-readable recording medium may include a hardware device configured to store and execute program instructions. For example, the computer-readable recording medium may be magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; ROM; RAM; flash memory, etc. The program instructions may include not only machine code, such as that generated by a compiler, but also high-level language code that can be executed by a computer through an interpreter, etc.

[0047] Hereinafter, with reference to FIG. 4, one of the operation methods of the object control data automatic generation system (100) will be described as an object control data automatic generation method.

[0048] FIG. 4 is a flowchart illustrating a method for automatically generating object control data according to an embodiment of the present invention.

[0049] Referring to FIG. 4, the method for automatically generating object control data according to one embodiment of the present invention consists of steps S210 to S250. The method for automatically generating object control data illustrated in FIG. 4 is according to one embodiment, and the steps of the method for automatically generating object control data according to the present invention are not limited to the embodiment illustrated in FIG. 4 and may be added, changed, or deleted as necessary.

[0050] Step S210 is the step of collecting sensor data from the controlled area.

[0051] The processor (110) of the object control data automatic generation system (100) collects sensor data (e.g., images, point clouds) of a control target area (e.g., a parking lot) using a sensor (150). The sensor (150) may be any one of a camera, radar, and lidar, or a combination thereof.

[0052] The processor (110) of the object control data automatic generation system (100) can periodically collect sensor data of the control target area. For example, the processor (110) can collect sensor data for two weeks at a interval of 10 minutes.

[0053] For example, the sensor (150) may be an IP camera that captures the control target area, which is the detection range of the target object. An IP camera refers to a digital video camera capable of transmitting collected video information using an internet protocol through a connected network. The processor (110) of the object control data automatic generation system (100) may collect snapshot images corresponding to the sensor data of the IP camera at regular intervals and store them in a memory (130) or a storage device (140). If the control target area is a parking lot, the processor (110) may collect sensor data of vehicles of various types and sizes, such as light cars, medium cars, trucks, and buses. In addition, considering various lighting and weather conditions, the processor (110) may obtain sensor data for various environments. That is, even within the same control target area, the location of the recognized object may differ depending on lighting conditions or weather, so the processor (110) may set the data collection period of the sensor (150) to a long period (e.g., one year). In this case, the sensor (150) collects sensor data regarding various lighting and weather conditions of the same controlled area and transmits it to the processor (110). For example, the processor (110) obtains sensor data for day and night when the controlled area is outdoors, and obtains sensor data regarding various illuminances when the controlled area is indoors. Additionally, the processor (110) obtains sensor data regarding weather changes such as clear, rain, and snow.

[0054] Meanwhile, in the sensor data collected by the sensor (150), it must be taken into account that the object to be detected (e.g., a vehicle) is obscured by a structure or other object. That is, the sensor (150) may collect sensor data in which only a part of the target object (e.g., a vehicle) is visible. The object detection model described below must be a model capable of handling such obscuration of the object.

[0055] FIGS. 5a to 5d are example drawings of sensor data collected through an IP camera. As shown in FIG. 5, the IP camera, which is a sensor (150) of the object control data automatic generation system (100), collects sensor data regarding various lighting conditions or weather. For example, FIG. 5d is an example drawing of sensor data in the case of snow. In the examples of FIGS. 5a to 5d, the obscuration status of objects (e.g., vehicles) or areas of interest (e.g., parking spaces) also varies.

[0056] Returning to Fig. 4, step S220 is explained.

[0057] Step S220 is the step of generating and storing object detection data.

[0058] The processor (110) generates object detection data using an artificial intelligence (e.g., deep learning)-based object detection model stored in memory (130) or storage device (140) based on sensor data collected in step S210, and accumulates and stores the object detection data in the database of the storage device (140).

[0059] FIGS. 6B, FIGS. 6C, and FIGS. 6D show the results of accumulating object detection data over a predetermined period, displayed on the image of FIG. 6A. In FIGS. 6B, FIGS. 6C, and FIGS. 6D, the object detection locations are represented by an intersection line, the center point of the object, and a bounding box, respectively.

[0060] Here, object detection data is data representing the location of an object (e.g., a vehicle), and may be data represented by two intersecting line segments (intersection lines, see FIG. 6b), a point representing the center location of the object (intersection point, see FIG. 6c), or a bounding box (see FIG. 6d). In this case, if the form of the object detection data is two intersecting line segments (intersection lines), the location of the detected object becomes the intersection point of the two line segments. The endpoints of the two line segments forming the intersection line correspond to the vertices of the bounding box, and the intersection point corresponds to the center of the bounding box. However, the representation of the object detection data is not limited to the examples described above (two line segments, bounding box), and there may be various other representation forms.

[0061] For example, when the sensor data is a video frame, the processor (110) stores one or more intersection lines, intersection points (center positions of objects), or bounding boxes for one or more objects on a frame-by-frame basis in the database of the storage device (140). At this time, the processor (110) can accumulate and store the detected positions or ranges of objects based on the positions of the pixels of the video frame.

[0062] In addition, as a countermeasure against the occlusion of an object or a region of interest, in an object control data automatic generation system (100) according to one embodiment of the present invention, the object detection model used by the processor (110) may be a model to which an amodal instance segmentation technique and / or a hierarchical occlusion modeling technique are applied. The amodal instance segmentation technique is a technique that predicts not only the visible part of an object but also the part that is obscured by other objects, and the hierarchical occlusion modeling technique is a technique that hierarchizes the occlusion relationships between objects.

[0063] Meanwhile, the object detection model may further include an object classification model. In this case, the object detection data may include not only the location of the object but also the type of the object. For example, the object classification model may classify the detected vehicle into light cars, medium cars, trucks, buses, etc.

[0064] Step S230 is the step of acquiring data for setting the region of interest.

[0065] Area of ​​Interest setting data is data that the object control data automatic generation system (100) uses as a standard to automatically generate an area of ​​interest (which may be referred to as 'area of ​​interest data'). That is, the area of ​​interest setting data is data that serves as a standard for generating an area of ​​interest (ROI), and includes the location and / or size of candidate areas of the area of ​​interest. For example, the area of ​​interest setting data may be data expressed as a baseline or a bounding box. However, the form of the area of ​​interest setting data of the present invention is not limited to the examples described above (baseline, bounding box), and the form of the area of ​​interest setting data may vary. When the area to be controlled is a parking lot, the area of ​​interest setting data may be the estimated location and / or range of a parking space.

[0066] For example, the processor (110) can identify an area where objects are mainly detected (e.g., an area where vehicles are mainly detected) based on accumulated object detection data, and generate area of ​​interest setting data (e.g., estimated location data of a parking space, estimated range data of a parking space) based on the area where objects are mainly detected.

[0067] As a specific example, if the area under control is a parking lot, the processor (110) can generate area of ​​interest setting data by tracking the location where the vehicle mainly stays based on object detection data for the vehicle and gradually determining the location of the parking space. The processor (110) generates object detection data by applying an object detection model to the vehicle detected in the video frames (sensor data) accumulated through the IP camera (this is the part described in step S220), and then, in step S230, can gradually determine the locations of the parking space by tracking the location where the vehicle mainly stayed based on the object detection data. That is, the processor (110) can generate area of ​​interest setting data (estimated location data of the parking space) using an artificial intelligence-based area of ​​interest setting model using the accumulated object detection data. The above area of ​​interest setting model may be any one of the clustering models, such as the K-Means clustering algorithm. The processor (110) can determine the center and size of the parking space based on the location where the vehicle is repeatedly parked. The processor (110) can automatically group object detection data by applying a clustering technique to the accumulated object detection data. For example, multiple bounding boxes generated as a result of detecting multiple vehicles parked in a specific parking space over a period of one week can be classified into one group through the application of a clustering model. The processor (110) generates region of interest setting data based on the classification results of the clustering model. The processor (110) can generate region of interest setting data by calculating the average of the object detection data included in each group.For example, the processor (110) can generate bounding boxes corresponding to each group by calculating the average of the center positions and sizes of multiple bounding boxes corresponding to one group and generating a new bounding box (a bounding box with the average of the center positions as the center and the average of the sizes as the size), and the bounding boxes generated corresponding to each group can be used as region of interest setting data. Even when the object detection data is an intersection line, the processor (110) can generate region of interest setting data composed of multiple intersection lines or bounding boxes in the same way as above. As another example, the processor (110) can generate a baseline by connecting the center positions of the bounding boxes (or intersection lines) corresponding to each group and set the baseline as region of interest setting data.

[0068] In an example different from the examples described above, the object control data automatic generation system (100) can receive interest area setting data from an administrator based on sensor data. For example, the administrator can input interest area setting data into the object control data automatic generation system (100) by drawing a baseline corresponding to the interest area setting data on a video frame.

[0069] Figures 7 and 8 are diagrams showing a baseline for object control and accumulated object detection data displayed on specific sensor data.

[0070] In Figures 7 and 8, CL1 and CL2 represent baselines which are region of interest setting data, and OD1 and OD2 represent object detection data (intersection lines) accumulated over a set period.

[0071] Returning to Fig. 4, steps S240 and below will be explained.

[0072] Step S240 is the step of creating a Region of Interest (ROI).

[0073] Additionally, among the area of ​​interest setting data, the area of ​​interest setting data corresponding to a location where the number of object detections calculated based on accumulated object detection data is less than a predetermined threshold may be a false detection or a non-parking area (where a vehicle is parked in a place other than a parking space). Accordingly, the processor (110) determines the area of ​​interest by excluding the area of ​​interest setting data corresponding to a location where the number of object detections is less than the threshold from the area of ​​interest setting data obtained in step S230. In other words, the processor (110) creates an area of ​​interest by selecting the area of ​​interest setting data where the number of object detections is greater than or equal to a predetermined threshold based on accumulated object detection data. This is because if an object (e.g., a vehicle) is repeatedly detected at the same location in multiple sensor data (e.g., video frames), the reliability of the area of ​​interest corresponding to that location will be high.

[0074] The shape of the region of interest generated in this step may vary. Depending on the settings, the processor (110) may generate a region of interest in the shape of a straight line (including a line segment) or a polygon. However, the present invention does not impose any restrictions on the shape of the region of interest.

[0075] FIGS. 9a to 9d are example drawings of regions of interest. FIGS. 9a and 9b are example drawings of a region of interest represented by a straight line (more precisely, a set of line segments), and FIGS. 9c and 9d are example drawings of a region of interest represented by a set of bounding boxes.

[0076] Meanwhile, after step S240, the processor (110) can determine how to perform control using the object detection data. Specifically, the processor (110) can determine which part of the object detection data to use as a basis for performing control.

[0077] For example, if the controlled area is a parking lot and the region of interest takes the form of a line segment, each line segment can correspond one-to-one with each parking space. Additionally, it is assumed that the controlled object detection data takes the form of the diagonal of the bounding box surrounding the detected object.

[0078] When the object control data automatic generation system (100) performs control regarding the parking status of a vehicle, the processor (110) can determine whether the vehicle is parked based on whether the line segment-shaped region of interest (ROI) intersects (meets) at least one of the two diagonals (top-left to bottom-right diagonal, top-right to bottom-left diagonal) of the bounding box, which is object detection data. Furthermore, the processor (110) can determine and set which of the two diagonals to apply for control for each region of interest (ROI, corresponding to a parking space) based on the accumulated object detection data. For example, the processor (110) can determine whether the vehicle is parked based on the top-right to bottom-left diagonal of the object detection data for parking space A corresponding to region of interest a (ROIa), and determine whether the vehicle is parked based on the top-left to bottom-right diagonal of the object detection data for parking space B corresponding to region of interest b (ROIb).

[0079] FIGS. 10a and FIGS. 10b are example drawings showing the diagonal components of detection data accumulated in sensor data collected through an IP camera. FIG. 10a shows the top-left to bottom-right diagonal (hereinafter 'top-left diagonal') among the accumulated detection data, and FIG. 10b shows the top-right to bottom-left diagonal (hereinafter 'top-right diagonal') among the accumulated detection data. When comparing FIG. 10a and FIG. 10b, it can be seen that the top-left diagonal of FIG. 10a distinguishes each parking space more clearly than the top-right diagonal of FIG. 10b. This is because the top-right diagonal of FIG. 10b overlaps relatively often with the top-right diagonal of adjacent parking spaces.

[0080] FIGS. 11a and 11b show line segments (hereinafter referred to as 'center-connecting line segments') connecting the centers of adjacent bounding boxes when the shape of the region of interest is a bounding box, as indicated in FIGS. 10a and 10b. In FIGS. 11a, a portion of the center-connecting line segments connecting the centers of adjacent regions of interest is indicated as CC1, and in FIGS. 11b, a portion of the center-connecting line segments connecting the centers of adjacent regions of interest is indicated as CC2. The size of the acute angle formed by the extension of the center-connecting line segment and the diagonal is larger in FIGS. 11a than in FIGS. 11b, and it can be seen that for the region of interest where the center-connecting line segment is indicated, it is more appropriate to perform control based on the upper-left diagonal.

[0081] Accordingly, the object control data automatic generation system (100) can increase calculation efficiency and control accuracy by selecting whether to determine whether each region of interest generated through step S240 intersects based on which direction diagonal (upper left diagonal or upper right diagonal).

[0082] If the shape of the region of interest is a line segment, the processor (110) can calculate the average of the acute angles formed by the line segment corresponding to the region of interest and the upper-left and upper-right diagonals of the accumulated object detection data, and then set the diagonal direction with the larger average of the acute angles as the diagonal direction that serves as the control standard.

[0083] Additionally, if the shape of the region of interest is a bounding box, the processor (110) can compare the average of the acute angles formed by the extension of the center connecting line segment and the diagonal of each direction, and set the diagonal direction with the larger average of the acute angles as the diagonal direction that serves as the control standard.

[0084] FIG. 12 is a diagram showing the center connecting line segments (Seg1, Seg2) corresponding to the area of ​​interest ROI2 when the shape of the area of ​​interest (ROI1, ROI2, ROI3) is a bounding box. To determine the diagonal direction to apply to ROI2, the processor (110) calculates the average of the acute angles formed by the extensions of the center connecting line segments Seg1 and Seg2 for each diagonal of the accumulated object detection data. Based on the average of the acute angles calculated for each diagonal, the processor (110) can determine the diagonal direction that serves as the control standard. For example, regarding the area of ​​interest ROI2, if the average of the acute angles formed when the upper-left diagonal is applied is greater than the average of the acute angles formed when the upper-right diagonal is applied, the processor (110) can determine the control standard diagonal direction for the area of ​​interest ROI2 as the upper-left diagonal.

[0085] FIG. 13 shows the extension of the line segment (Seg1 in FIG. 12) connecting the centers of adjacent regions of interest (XSeg1, extending, for example, to the corner of the bounding box which is the region of interest) and the upper-left diagonal of the object detection data (D TL A drawing showing the angles (θ1, θ2) formed by ) and the processor (110) uses the average of the acute angle θ1 to determine the direction of the diagonal that serves as the control reference.

[0086] As an example different from the method described above, the processor (110) may determine the direction of the diagonal serving as the control standard based on the distance (e.g., minimum distance between line segments) between the diagonal of the object detection data in a specific direction for the area of ​​interest to be determined and the diagonal of the object detection data in the same direction for the adjacent area of ​​interest. This takes into account that in FIG. 10a, the upper-left diagonals of the object detection data between adjacent areas of interest are separated by a certain distance from each other, whereas in FIG. 10b, the upper-right diagonals of the object detection data between adjacent areas of interest meet each other or are relatively close in distance.

[0087] Figure 14 is a diagram illustrating a case where the diagonal of the object detection data serving as the control standard is set differently depending on the area of ​​interest.

[0088] In FIG. 14, the region of interest ROIa is shown in blue, and the region of interest ROIb is shown in red. For example, the processor (110) can determine whether an object exists based on the upper right diagonal for the region of interest (ROIa) shown in blue, and whether an object exists based on the upper left diagonal for the region of interest (ROIb) shown in red.

[0089] Returning to Fig. 4, step S250 is explained.

[0090] Step S250 is the step of generating object control data.

[0091] Once the area of ​​interest is determined, the object control data automatic generation system (100) automatically assigns a management number (e.g., parking space number) to each area of ​​interest and generates object control data by monitoring the control target area in real time.

[0092] The processor (110) generates object control data (which may be abbreviated as 'control data') for a control target area based on object detection data corresponding to recent sensor data among the sensor data collected by the sensor (150) and the region of interest generated in step S240. If the components of the object detection data to be applied to each region of interest (e.g., diagonal direction - top-left diagonal, top-right diagonal) are set, the processor (110) can generate control data based on whether the components of the object detection data intersect the region of interest. For example, the processor (110) can generate control data indicating that an object exists in the region of interest if the top-left diagonal among the object detection data having a diagonal shape intersects the region of interest having a line segment shape. In addition, as another example, the processor (110) can generate control data based on whether the top-left diagonal among the object detection data having a diagonal shape intersects the midpoint connecting line of two opposite sides of the region of interest having a bounding box shape.

[0093] When the control target area of ​​the object control data automatic generation system (100) is a parking lot and the area of ​​interest is a parking space, the object control data generated by the processor (110) may include data indicating whether a vehicle is parked in a parking space (each parking space corresponds to a management number). The object control data may further include the type of vehicle parked in the parking space.

[0094] Additionally, the processor (110) can generate statistical data regarding an area of ​​interest based on object control data accumulated over a certain period. For example, the processor (110) can calculate the probability distribution of the average occupancy rate of a parking lot on a specific date or during a specific period, or the occupancy rate by time of day during a specific period, and can generate time-series data of the parking lot occupancy rate. Additionally, the processor (110) can generate data comparing the parking lot occupancy rates on weekdays and weekends.

[0095] FIG. 15 is an example drawing of a region of interest and control data generated by the method for automatically generating object control data according to the present invention. In the example of FIG. 15, a region of interest (PS1, PS2) automatically generated by an object control data automatic generation system (100) based on region of interest setting data, and control data (CD1, CD2) indicating whether a vehicle is parked are shown. In FIG. 15, the object control data automatic generation system (100) can display a parking space that is parked and an empty parking space in different colors (CD1 is yellow, CD2 is red). In FIG. 15, for the region of interest on the right, control data was generated based on the upper right diagonal among the object detection data, and for the region of interest on the left, control data was generated based on the upper left diagonal among the object detection data.

[0096] Figures 16a and 16b are examples of real-time object control screens, in which a management number is assigned to the area of ​​interest corresponding to each parking space, and the area of ​​interest that is currently parked and the area of ​​interest that is empty are each indicated by circles of different colors.

[0097] The above-described method for automatically generating object control data has been explained with reference to the flowchart presented in the drawings. For simplicity of explanation, the method has been illustrated and described in a series of blocks; however, the present invention is not limited to the order of said blocks, and some blocks may occur in a different order or simultaneously with other blocks as illustrated and described herein, and various other branches, flow paths, and sequences of blocks that achieve the same or similar results may be implemented. Furthermore, not all illustrated blocks may be required for the implementation of the method described herein.

[0098] Meanwhile, in the description with reference to FIG. 4, each step may be further divided into additional steps or combined into fewer steps according to an embodiment of the present invention. Also, some steps may be omitted as necessary, and the order between steps may be changed. Furthermore, even if other omitted details are included, the contents of FIG. 3 may be applied to the contents of FIG. 4. Also, the contents of FIG. 4 through FIG. 10 may be applied to the contents of FIG. 3.

[0099] Although the present invention has been described above with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as described in the following claims. Explanation of the symbols

[0100] 100: Object Control Data Automatic Generation System 110: Processor 120: Communication device 130: Memory 140: Storage device 150: Sensor 160: Bus

Claims

Claim 1 A method for automatically generating object control data, comprising: a step of collecting sensor data of a control target area using a sensor; a step of the system generating object detection data using an artificial intelligence-based object detection model based on the sensor data and storing it in a database; a step of the system generating interest area setting data based on the object detection data; and a step of the system generating an interest area by removing a portion of the interest area setting data based on the object detection data. Claim 2 A method for automatically generating object control data according to claim 1, further comprising the step of the system generating object control data for the control target area based on object detection data corresponding to recent sensor data among the sensor data and the region of interest. Claim 3 In claim 1, the sensor is one of a camera, radar, and lidar, a method for automatically generating object control data. Claim 4 A method for automatically generating object control data according to claim 1, wherein the object detection data is data represented by two intersecting line segments, and the location of the detected object is the intersection point of the two line segments. Claim 5 In claim 1, the above-mentioned interest region setting data is a method for automatically generating object control data, which is data represented by a baseline. Claim 6 A method for automatically generating object control data according to paragraph 2, wherein the control target area is a parking lot, the area of ​​interest is a parking space, and the object control data is data indicating whether a vehicle is parked in the parking space. Claim 7 A method for automatically generating object control data according to claim 1, wherein the step of generating the interest area setting data comprises the system applying a clustering model to the object detection data collected over a predetermined period to classify the object detection data into a plurality of groups, and generating the interest area setting data based on the object detection data included in each group. Claim 8 An object control data automatic generation system comprising: a memory for storing computer-readable commands; and at least one processor implemented to execute said commands, wherein the at least one processor is configured to collect sensor data of a control target area by executing said commands, generate object detection data using an artificial intelligence-based object detection model based on said sensor data and store it in a database, generate interest area setting data based on said object detection data, and generate an interest area by removing a portion of said interest area setting data based on said object detection data. Claim 9 An object control data automatic generation system according to claim 8, wherein at least one processor is configured to generate object control data for the control target area based on object detection data corresponding to recent sensor data among the sensor data and the region of interest. Claim 10 In paragraph 8, the sensor is an object control data automatic generation system that is one of a camera, radar, and lidar. Claim 11 In claim 8, the object detection data is data represented by two intersecting line segments, and the location of the detected object is the intersection point of the two line segments, an object control data automatic generation system. Claim 12 In paragraph 8, the above-mentioned interest area setting data is an object control data automatic generation system, which is data represented by a baseline. Claim 13 In claim 9, the above-mentioned control target area is a parking lot, the above-mentioned area of ​​interest is a parking space, and the above-mentioned object control data is data indicating whether a vehicle is parked in the above-mentioned parking space, an object control data automatic generation system. Claim 14 An object control data automatic generation system according to claim 8, wherein at least one processor is configured to apply a clustering model to the object detection data collected over a predetermined period to classify the object detection data into a plurality of groups, and to generate the area of ​​interest setting data based on the object detection data included in each group.

Images