Method and system for preventing illegal parking in electric vehicle charging areas using a deep learning-based optical character recognition engine

Abstract

An embodiment may provide an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine, wherein the device is installed within a parking area for parking a vehicle and is located adjacent to a speed bump within the parking area; and a server that recognizes a vehicle number within a captured image based on the analysis of a captured image received from the device and transmits the result to the device. The device comprises: a main body; a rotating bar connected to a rotating shaft on a rotating bar receiving groove of the main body and rotatable within a predetermined angle range; a controller that controls the rotation of the rotating bar based on the recognition result of the vehicle number; and a sensing unit for detecting the vehicle and capturing the vehicle. The main body is installed in the parking area, and the rotating bar is installed in the main body and is installed vertically from the main body, and is rotatable towards the rear of the main body around the connection point between the main body and the rotating bar.

Description

Technology Field

[0001] The present invention relates to a method for preventing illegal parking in electric vehicle charging areas using a deep learning-based optical character recognition engine and a system for the same. Background Technology

[0002] The 21st century is an era where environmental protection for sustainable development has emerged as a critical social task, leading to the rapid implementation of policies focused on environmental protection. Influenced by these trends, the adoption of electric vehicles (EVs) in the automotive market is accelerating, and various policies are being implemented to facilitate this. A representative example is the installation and operation of dedicated EV charging zones in parking lots. However, the problem is that despite these efforts, a significant number of EV drivers remain dissatisfied with the charging infrastructure, and related complaints are growing due to instances of internal combustion engine vehicles illegally parking in these charging areas. Despite these significant issues, current response measures are limited to the imposition of fines; since action is only taken after someone reports illegal parking that has already occurred, there are limitations in preventing the fundamental damage caused by such parking. To address this problem, an illegal parking warning and prevention system has been proposed that utilizes a real-time object detection model to recognize the dedicated logo included on the license plates of electric vehicles. Specifically, a warning system is being introduced that uses YOLO, a deep learning-based real-time object detection model, to learn and recognize the dedicated mark included on electric vehicle license plates, sounding an alarm if the mark is not confirmed. Furthermore, a system has been proposed that adds a hardware device with license plate storage and parking blocking functions to the above method. However, there is a concern that this technology may face issues due to the forgery of the dedicated logo or problems during the recognition process under various environmental conditions. Additionally, it has been pointed out that the system cannot be applied to vehicles that lack the dedicated mark, such as electric taxis or plug-in hybrid vehicles, even though they are eligible to use charging areas. Prior art literature

[0003] Registered Patent KR 10-1538592 Registered Patent KR 01752346 Registered Patent KR 10-2023-0119396 Registered Patent KR 101892565 The problem to be solved

[0004] The present invention provides a method for preventing illegal parking of electric vehicle charging areas using a deep learning-based optical character recognition engine and a system for the same, which blocks access to a charging area when a vehicle approaching the charging area is an internal combustion engine vehicle that is not qualified to use the charging area.

[0005] In addition, the present invention provides a method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine and a system for the same, which extracts a vehicle number using optical character recognition (OCR) when a vehicle approaches the charging area, checks whether the vehicle is qualified to use the charging area by querying the fuel type information of the vehicle registered with the extracted number, and permits access to the charging area if the vehicle is qualified.

[0006] In addition, the present invention provides a method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine and a system for the same, which can quickly resolve the situation when a vehicle that is not allowed to park attempts to forcibly park in a parking area. means of solving the problem

[0007] An embodiment may provide an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine, wherein the device is installed within a parking area for parking a vehicle and is located adjacent to a speed bump within the parking area; and a server that recognizes a vehicle number within a captured image based on the analysis of a captured image received from the device and transmits the result to the device. The device comprises: a main body; a rotating bar connected to a rotating shaft on a rotating bar receiving groove of the main body and rotatable within a predetermined angle range; a controller that controls the rotation of the rotating bar based on the recognition result of the vehicle number; and a sensing unit for detecting the vehicle and capturing the vehicle. The main body is installed in the parking area, and the rotating bar is installed in the main body and is installed vertically from the main body, and is rotatable towards the rear of the main body around the connection point between the main body and the rotating bar.

[0008] In another aspect, the sensing unit may include an illuminance sensor for measuring illuminance, and the opening and closing device may further include an acoustic output unit for outputting a warning message regarding illegal parking in the parking area, and a light output unit for outputting light when the illuminance value sensed by the illuminance sensor is less than a preset value, thereby providing an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine.

[0009] In another aspect, the above-described opening and closing device may provide an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine, which further includes a main body rotation part installed on the floor of the parking area to rotate the main body part parallel to the ground.

[0010] In another aspect, the controller can provide an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine that maintains the initial position of the rotating bar when the result of recognizing the license plate of the vehicle entering the parking area determines that it is illegal parking, and rotates the rotating bar toward the rear side of the main body when it determines that it is normal parking.

[0011] In another aspect, the controller may provide an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine that maintains the initial position of the rotating bar when it determines that the vehicle entering the parking area is illegally parked based on the recognition result of the vehicle number of the vehicle, and detects the position of the vehicle in real time to rotate the main body so that the front of the rotating bar and the rear or front of the vehicle face each other.

[0012] In another aspect, an electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine can be provided, wherein the controller determines that the vehicle entering the parking area is illegally parked based on the recognition result of the vehicle number of the vehicle, and maintains the initial position of the rotating bar, and the rotating bar rotates when the vehicle presses the rotating bar.

[0013] In another aspect, in a system for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine, the opening and closing device detects a vehicle entering the parking area, generates a captured image of the vehicle, and transmits the captured image to a server; the server detects a vehicle license plate area within the captured image, recognizes the vehicle number within the vehicle license plate area, and then looks up the vehicle number; and the opening and closing device may provide a method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine by maintaining the initial position of the rotating bar or rotating the rotating bar at a predetermined angle based on the result of looking up the vehicle number.

[0014] In another aspect, the server can provide a method for preventing illegal parking in an electric vehicle charging area by utilizing a deep learning-based optical character recognition engine that recognizes the vehicle number by determining the vehicle number within the vehicle number plate area as one of a plurality of types, recognizing the number only up to the number of characters matched to the determined type, and deleting the characters recognized outside the matched number. Effects of the invention

[0015] The embodiment may provide a method for preventing illegal parking of electric vehicle charging areas and a system for the same using a deep learning-based optical character recognition engine that blocks access to the charging area when the vehicle approaching the charging area is an internal combustion engine vehicle that is not qualified to use the charging area.

[0016] In addition, the embodiment provides a method for preventing illegal parking of electric vehicle charging areas using a deep learning-based optical character recognition engine and a system for the same, which extracts a vehicle number using optical character recognition (OCR) when a vehicle approaches a charging area, checks whether the vehicle is qualified to use the charging area by querying the fuel type information of the vehicle registered with the extracted number, and permits access to the charging area if the vehicle is qualified.

[0017] In addition, the embodiment can prevent illegal parking by outputting a warning message regarding illegal parking to illegally parked vehicles.

[0018] In addition, the embodiment installs an opening and closing device within a parking area for a single vehicle, allowing each parking area, such as a disabled parking area or an electric vehicle-only parking area, to be managed independently of each other for illegal parking.

[0019] In addition, the embodiment can prevent the opening and closing device from being damaged even if the opening and closing device collides with the vehicle in a situation where an unauthorized vehicle is illegally parked. Brief explanation of the drawing

[0020] FIG. 1 is a system for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine according to an embodiment of the present invention. Figure 2 is a block diagram of a switching device. Figure 3 is a perspective view of the opening and closing device. Figure 4 illustrates the rotation of the rotating bar of the opening and closing device. Figure 5 schematically illustrates a vehicle entering a parking area equipped with an opening and closing device. Figure 6 schematically illustrates the rotation of a rotating bar by physical contact with a vehicle. Figure 7 schematically illustrates a rotating bar repeatedly rotating in the forward and reverse directions to apply physical impact to a vehicle. FIG. 8 is a perspective view of an opening and closing device according to various embodiments of the present invention. FIG. 9 is a perspective view of an opening and closing device according to various embodiments of the present invention. FIG. 10 is a perspective view of an opening and closing device with the distance detection unit in an open state. Figure 11 illustrates an exemplary structure of a target prediction model used in performing a target prediction method in an artificial intelligence model. Specific details for implementing the invention

[0021] The present invention is capable of various modifications and may have various embodiments; therefore, specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present invention, and the methods for achieving them, will become clear by referring to the embodiments described in detail below together with the drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various forms. In the following embodiments, terms such as "first," "second," etc., are used not in a limiting sense but for the purpose of distinguishing one component from another. Furthermore, singular expressions include plural expressions unless the context clearly indicates otherwise. Also, terms such as "include" or "have" mean that the features or components described in the specification exist, and do not preclude the possibility that one or more other features or components may be added. Additionally, in the drawings, the size of components may be exaggerated or reduced for convenience of explanation. For example, the size and thickness of each component shown in the drawings are arbitrarily depicted for convenience of explanation, so the present invention is not necessarily limited to what is illustrated.

[0022] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. When describing with reference to the drawings, identical or corresponding components are given the same reference numerals, and redundant descriptions thereof will be omitted.

[0023] FIG. 1 is a system for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine according to an embodiment of the present invention. FIG. 2 is a block diagram of an opening / closing device, FIG. 3 is a perspective view of the opening / closing device, and FIG. 4 illustrates the rotation of the rotating bar of the opening / closing device. FIG. 5 schematically illustrates a vehicle entering a parking area where the opening / closing device is installed, FIG. 6 schematically illustrates the rotating bar rotating due to physical contact with the vehicle, and FIG. 7 schematically illustrates the rotating bar repeatedly rotating in the forward and reverse directions to apply physical impact to the vehicle.

[0024] Referring to FIGS. 1 and 5, an electric vehicle charging area illegal parking prevention system (10) utilizing a deep learning-based optical character recognition engine according to an embodiment of the present invention may include an opening / closing device (100) and a server (200).

[0025] A network refers to a connection structure capable of exchanging information between each node, such as a switch (100) and a server (200). Examples of such networks include, but are not limited to, a 3GPP (3rd Generation Partnership Project) network, an LTE (Long Term Evolution) network, a WIMAX (World Interoperability for Microwave Access) network, the Internet, a LAN (Local Area Network), a Wireless LAN (Wireless Local Area Network), a WAN (Wide Area Network), a PAN (Personal Area Network), a Bluetooth network, a satellite broadcasting network, an analog broadcasting network, and a DMB (Digital Multimedia Broadcasting) network.

[0026] The opening and closing device (100) may be installed within a parking area. Here, the parking area is defined as an area for parking a single vehicle.

[0027] The opening / closing device (100) can generate a captured image of the vehicle (20) when it senses the entry of the vehicle (20) into a preset parking area. The opening / closing device (100) can then transmit the captured image to a server (200). In various embodiments, the opening / closing device (100) may generate a plurality of captured images of the vehicle (20) according to a preset time cycle and analyze the plurality of captured images.

[0028] The server (200) can analyze the received captured image. The server (200) can detect a license plate area within the captured image and detect a vehicle number within the detected license plate area. The server (200) can look up the recognized vehicle number in a pre-registered database. In some embodiments, the server (200) can communicate with another external server to request a lookup of the recognized vehicle number. The server (200) can transmit the result information of the vehicle number lookup to the opening / closing device (100). The server (200) can recognize the vehicle number within the license plate area based on a deep learning model. The deep learning model may be EasyOCR or Pororo (Platform of neural models for natural language processing), but is not limited thereto. In various embodiments, the server (200) determines one of a plurality of preset types based on the length of the vehicle number recognized within the license plate area, recognizes the number only up to the number of characters matched to the determined type, and deletes the characters recognized outside the number of characters matched.

[0029] In detail, the server (200) can designate a vehicle number recognized within the license plate area as either a first or second type. A Korean vehicle license plate may consist of two numbers, one letter, and four numbers, or three numbers, one letter, and four numbers. The former can be defined as the first type, and the latter as the second type.

[0030] The server (200) can check if the length of the characters is seven when the pattern of the extracted number starts with the first character and corresponds to the first type, and if it corresponds to the second type, it can check if the length of the characters is eight.

[0031] The server (200) can recognize seven characters when recognizing the vehicle number as a first type and delete the characters recognized thereafter, and when recognizing it as a second type, recognize eight characters and delete the characters recognized thereafter, thereby improving the accuracy of vehicle number recognition.

[0032] The opening / closing device (100) may include a controller (110). The controller (110) may include a processor (111), a memory (112), a communication unit (113), and a motor unit (114). Additionally, the opening / closing device (100) may include a main body unit (120) incorporating the controller (110), a rotating bar (130), and a sensing unit (140). In various embodiments, the opening / closing device (100) may further include an output unit (150).

[0033] The rotating bar (130) is installed on the main body (120). The rotating bar (130) can be installed vertically from the main body (120) installed on the floor of the parking area. The lower portion of the rotating bar (130) is inserted into the rotating bar receiving groove (121) of the main body (120), and the lower portion of the rotating bar (130) can be installed on the rotating shaft (131) which is installed on the main body (120) and has a portion exposed through the rotating bar receiving groove (121).

[0034] The rotating bar (130) may be configured as a bar type protruding from the main body (120) at a predetermined height. The rotating bar (130) may be configured to rotate within a predetermined angle range around a rotation axis centered at the connection point with the main body (120).

[0035] The processor (111) can control the overall operation of each device and perform data processing for a series of operations to be described later. The processor may be ASICs (application specific integrated circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), controllers, microcontrollers, microprocessors, or any other type of processor for performing functions.

[0036] The memory (112) can store instructions and data for controlling the operation of the processor. The memory may include instructions that operate as a series of processes. Additionally, the memory (112) may include various instructions for executing a pre-trained artificial intelligence model that the processor uses to determine the location of the vehicle. The memory (112) may be various storage devices such as ROM, RAM, EPROM, flash drive, hard drive, etc. However, it is not limited thereto, and the memory (112) may be web storage that performs the storage function of the memory on the internet.

[0037] The communication unit (113) can wirelessly transmit and receive data with at least one of a base station, an external terminal, or any server on a mobile communication network built through a communication device capable of performing technical standards for mobile communication or communication methods (e.g., LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), 5G NR (New Radio), WIFI) or short-range communication methods.

[0038] The sensing unit (140) may include an image sensor (141), an illuminance sensor (142), and a proximity sensor (143).

[0039] The output unit (150) may include an audio output unit (151) and an optical output unit (152).

[0040] The controller (110) can obtain location information of the vehicle (2) by analyzing the captured image generated from the image sensor (141).

[0041] The controller (110) transmits a captured image generated from an image sensor (141) to a server (200) to enable the server (200) to recognize a vehicle number within the captured image, and can receive the result of the vehicle number recognition from the server (200).

[0042] The controller (110) can analyze the illuminance sensing information within the parking area of ​​the illuminance sensor (142) to determine whether the illuminance is below a preset value.

[0043] The controller (110) can determine whether an object is detected within the parking area by analyzing the sensing information of the proximity sensor (143).

[0044] The controller (110) can determine whether a vehicle is detected in the parking area by analyzing a captured image from an image sensor (141) when an object is detected in the parking area based on the sensing information of the proximity sensor (143).

[0045] The controller (110) receives a vehicle number recognition result from the server (200) and, if the vehicle (20) that has entered the parking area is determined to be a vehicle (20) that cannot be parked in the parking area, can output a warning message through the sound output unit (151).

[0046] The controller (110) can cause light to be output from the light output unit (152) when the illuminance value sensed by the illuminance sensor (142) is less than a preset value, and can control the image sensor (141) when light is output from the light output unit (152) to generate a captured image of an object within the parking area.

[0047] The controller (110) controls the motor unit (114) so ​​that the rotating bar (130) rotates within a predetermined angle range on the main body unit (120).

[0048] The main body (120) may be installed within a parking area. Within the parking area, a barrier (1) may be installed for the purpose of preventing the vehicle (20) from being parked outside the parking area and preventing damage to the vehicle (20). The opening / closing device (100) may be installed adjacent to the barrier (1). That is, the main body (120) may be installed adjacent to the barrier (1), and the rotating bar (130) may be configured to be rotatable within a predetermined angle range on the main body (120).

[0049] When the controller (110) receives a vehicle number recognition result from the server (200) and determines that the vehicle (20) that has entered the parking area can be parked within the parking area, it can rotate the rotating bar (130) toward the rear side of the main body (120) around the connection point between the rotating bar (130) and the main body (120). For example, the rotating bar (130) can be rotated from its initial position toward the rear side of the main body (120) at a maximum angle of 90 degrees. When the vehicle (20) is successfully parked within the parking area as a result of the rotation of the rotating bar (130), the rotating bar (130), including the main body (120), does not come into physical contact with the vehicle (20).

[0050] The controller (110) receives a vehicle number recognition result from the server (200) and, if it determines that a vehicle (20) that has entered the parking area cannot be parked within the parking area, maintains the initial position of the rotating bar (130) to allow the vehicle (20) to fully enter the parking area. Thus, illegal parking of the vehicle (20) can be prevented.

[0051] When the controller (110) determines that the vehicle (20) has left the parking area based on the sensing result of the sensing unit (140), it can control the motor unit (114) to rotate the rotating bar (130) so that the rotating bar (130) moves to an initial position.

[0052] A method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine according to an embodiment is as follows: the opening / closing device (100) recognizes a vehicle (20) approaching the parking area, transmits a captured image of the vehicle (20) to a server (200), the server (200) recognizes the vehicle number on the license plate in the captured image, the server (200) looks up the vehicle number and transmits the result to the opening / closing device (100), and the opening / closing device (100) can maintain the position of the rotating bar (130) at an initial position or rotate it by a predetermined angle based on the result of looking up the vehicle number.

[0053] Referring to FIGS. 6 and FIGS. 7, in various embodiments, when a vehicle (20) that has entered a parking area is determined to be unable to park within the parking area and the rotating bar (130) maintains its initial position, the vehicle (20) may physically collide with the rotating bar (130), and the rotating bar (130) may be configured to rotate in response to the distance traveled by the vehicle (20) by the pressure applied by the vehicle (20).

[0054] After the controller (110) determines that a vehicle (20) that has entered the parking area cannot be parked within the parking area, if the rotation of the motor unit (114) is detected in accordance with the rotation of the rotating bar (130), the rotating bar (130) can be rotated to a maximum angle. Then, the rotating bar (130) can deliver a warning message by applying physical impact to the vehicle by repeating forward and reverse rotations several times within a predetermined angle range. Through this, the driver of the illegally parked vehicle (20) can be induced to move the vehicle (20) out of the parking area.

[0055] FIG. 8 is a perspective view of an opening and closing device according to various embodiments of the present invention.

[0056] Referring to FIG. 8, the opening and closing device (100) according to various embodiments of the present invention may further include a main body rotation part (170). The main body rotation part (170) is installed within a parking area, and the main body part (120) is installed on the main body rotation part (170) and may be configured to rotate within a predetermined angle range while maintaining a horizontal position with respect to the ground with the main body rotation part (170) as the center.

[0057] After detecting the entry of the vehicle (20) into the parking area, the controller (110) can generate multiple captured images of the vehicle (20). At least one of the multiple captured images can be transmitted to the server (200) to enable the server (200) to recognize the vehicle number. Additionally, the controller (110) can detect the location of the vehicle in real time by analyzing at least some of the multiple captured images. Specifically, the controller (110) can continuously detect the direction of the vehicle (20) when the vehicle (20) enters the parking area.

[0058] Based on the result of the vehicle number lookup from the server (200), it can determine whether the vehicle (20) can be parked within the corresponding parking area. After determining that the vehicle (20) cannot be parked within the parking area, the controller (110) can control the main body rotation unit (170) to rotate the main body (120) based on the movement direction for parking of the vehicle (20) analyzed according to the analysis of the captured image. At this time, the controller (110) can control the main body rotation unit (170) so that the front of the rotating bar (130) facing the license plate on the front or rear of the vehicle (20) faces the license plate. Specifically, the controller (110) can rotate the main body (120) in real time so that the front of the license plate and the front of the rotating bar (130) remain parallel to each other. After the vehicle (20) has occupied a predetermined area within the parking zone, when the vehicle (20) comes into contact with the rotating bar (130), the front of the rotating bar (130) faces the front or rear of the vehicle (20). Specifically, when the vehicle (20) comes into contact with the rotating bar (130), the front of the rotating bar (130) may be parallel to the front or rear of the vehicle (20), or a virtual plane regarding the front of the rotating bar (130) and a virtual plane including a portion of the front or rear of the vehicle (20) that comes into contact with the front of the rotating bar (130) may be parallel to each other, or the acute angle formed by the normals of these virtual planes may be within a predetermined angle range. The vehicle (20) collides with the rotating bar (130), and as the vehicle (20) moves further into the parking zone, it presses the rotating bar (130), causing the rotating bar (130) to rotate forcibly. In addition, when a vehicle (20) parked illegally is not positioned normally within the parking area but moves obliquely, such as by part of the vehicle (20) moving out of the parking area, the side of the rotating bar (130) is pressed to prevent the direction of the force applied to the rotating bar (130) from being misaligned with the rotation axis of the rotating bar (130), thereby preventing the rotating bar (130) from being damaged.

[0059] FIG. 9 is a perspective view of an opening and closing device according to various embodiments of the present invention, and FIG. 10 is a perspective view of an opening and closing device in a state where the distance detection unit is open.

[0060] Referring to FIGS. 9 and 10, distance detection units (180) may be installed on both sides of a rotating bar (130) of a switching device (100) constituting various embodiments of the present invention. The distance detection units (180) may include first and second distance detection units (181, 182). The first distance detection unit (181) may be installed on one side of the rotating bar (130) and a portion of its area may be inserted into the interior of the rotating bar (130), and the second distance detection unit (182) may be installed on the other side of the rotating bar (130) and a portion of its area may be inserted into the interior of the rotating bar (130).

[0061] The distance detection unit (180) can be opened by rotating around the connection point with the rotating bar (130). When the distance detection unit (180) is opened, the rotating bar (130) and the distance detection unit (180) can have an overall T-shape.

[0062] A distance measuring sensor may be installed in the distance detection unit (180). Additionally, a plurality of distance measuring sensors may be installed spaced apart from each other in each of the first and second distance detection units (181, 182). The distance measuring sensors may be installed at a predetermined distance from each other on the front of the first and second distance detection units (181, 182) which form a surface parallel to the front of the rotating bar (130).

[0063] The controller (110) determines whether the vehicle (20) can be parked within the corresponding parking area based on the result of looking up the vehicle number from the server (200), and after determining that the vehicle (20) cannot be parked within the parking area, the controller (110) can determine the location of the vehicle (20) based on the analysis of the captured image of the vehicle. Then, the controller (110) can open the distance detection unit (180) and, based on the distance measurement detection result from the distance measurement sensors, control the main body rotation unit (170) so that the front of the rotating bar (130) and the license plate of the front of the vehicle (20) (vehicle parked in the front) / the license plate of the rear of the vehicle (20) (vehicle parked in the rear) are defined as a virtual plane, and a virtual vertical line perpendicular to the normal to the virtual plane is parallel to each other, thereby causing the rotating bar (130) to rotate. Generally, since the front or rear of the vehicle (20) has a curved surface, the rotating bar (130) can be controlled so that the license plate of the vehicle (20) and the front of the rotating bar (130) are parallel (considering a predetermined error) based on the analysis of the captured image and the distance measurement detection result of the distance detection unit (180), and accordingly, damage to the rotating bar (130) due to the mismatch between the impact point of the rotating bar (130) and the rotation direction of the rotating bar (130) due to the oblique parking of the vehicle (20) can be prevented.

[0064] Figure 11 illustrates an exemplary structure of a target prediction model used in performing a target prediction method in an artificial intelligence model.

[0065] In various embodiments, the processor (111) can detect the location of the vehicle based on an artificial intelligence model. The artificial intelligence model here may be a target prediction model.

[0066] The target prediction model utilized by the processor (111) to perform the target prediction method may include a structure in which an encoder and a decoder are connected. The target prediction model may include a structure similar to a Transformer model. For example, the target prediction model may include a structure similar to a Transformer model in which multiple attention operation layers are stacked to learn the importance patterns of multiple factor characteristic sequences of multivariate time series data. A Transformer model is an artificial intelligence model that learns context and meaning by tracking relationships within sequential data, such as words in a sentence, and is a model that performs predictions for target data based on input data by utilizing an encoder-decoder structure. The Transformer model includes an encoder-decoder structure in which an input sequence is received by an encoder and an output sequence is output by a decoder, and may include a structure in which N encoder-decoder structures are formed. Unlike the existing Transformer model, the target prediction model according to one embodiment By performing distributed-lag embedding on multivariate time series data and using a model with an interpretable multi-head attention layer applied to the encoder and decoder, it is possible to explain the influence of each of the multiple factor feature sequences included in the multivariate time series data on target prediction at various past time points.

[0067] The AI ​​model can determine the location of a target vehicle based on multivariate time-series data. The multivariate time-series data may include vehicle image data and distance information data. The AI ​​model can generate distributed-lag data by performing distributed-lag embedding on the multivariate time-series data, generate distributed-lag context data by performing encoding based on an interpretable multi-head self-attention mechanism on the distributed-lag data, generate target input data by performing token embedding and location embedding on the target data, and generate result data regarding the vehicle's location, the prediction result, and the explanation for the prediction result by performing decoding based on an interpretable multi-head cross-attention mechanism on the distributed-lag context data and the target input data.

[0068] Additionally, a factor feature vector can be generated by individually performing embeddings on each of the multiple lag sequences of each of the multiple factor feature sequences of multivariate time series data, a Global sequence position embedding (GSPE) vector for the factor feature vector, and a Local sequence position embedding (LSPE) vector for the factor feature vector, and lag distribution data can be generated by summing the aforementioned factor feature vector, GSPE vector, and LSPE vector. Furthermore, the GSPE vector may include different position information for an arbitrary first lag sequence of an arbitrary first factor feature sequence among the multiple factor feature sequences and an arbitrary second lag sequence of an arbitrary second factor feature sequence that is different from it, and the LSPE vector may include the same position information for each of the multiple factor feature sequences. Furthermore, the LSPE vector may include a structure in which multiple sub-vectors are connected, each containing position information for each of the multiple factor feature sequences, and the position information for each of the multiple factor feature sequences included in each of the multiple sub-vectors may be identical to each other.

[0069] Additionally, to generate differential distribution context data, a first attention operation is performed on differential distribution data based on a first interpretable multi-head self-attention layer having different queries and keys for multiple heads and the same value for all multiple heads; differential distribution data is summed with the data obtained through the first attention operation and normalization is performed to generate first normalized data; a first FFN operation is performed on the first normalized data based on a first feed forward network (FFN) layer; and differential distribution context data is generated by summing the first normalized data with the data obtained through the first FFN operation and performing normalization.

[0070] In addition, when generating target input data, masking processing can be performed first before performing token embedding and position embedding on the target data, and then token embedding and position embedding can be performed on the masked data to generate target input data.

[0071] In addition, in generating result data, a second attention operation is performed on target input data based on a second interpretable multi-head self-attention layer having different queries and keys for multiple heads and the same value for all multiple heads, and the target input data is summed with the data obtained through the second attention operation and normalization is performed to generate third normalized data, a third attention operation is performed on disparate distribution context data and third normalized data based on a first interpretable multi-head cross-attention layer having different queries and keys for multiple heads and the same value for all multiple heads, and the third normalized data is summed with the data obtained through the third attention operation and normalization is performed to generate fourth normalized data, a second FFN operation is performed on the fourth normalized data based on a second FFN layer, and the fourth normalized data is summed with the data obtained through the second FFN operation and normalization is performed to generate result data, and a predetermined linear operation is performed on the result data to generate final result data.

[0072] In addition, when performing a third attention operation based on a first interpretable multi-head cross-attention layer and generating fourth normalized data, prediction data related to the prediction result and explanation data related to the explanation of the prediction result may be generated.

[0073] Although the embodiments of the present invention have been described with a focus on electric vehicles, the present invention may also be applied to prevent illegal parking within designated parking areas for the disabled by looking up the license plate numbers of vehicles registered for the disabled.

[0074] The embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the computer-readable recording medium may be those specifically designed and configured for the present invention or those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include 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; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. Hardware devices may be modified into one or more software modules to perform processing according to the present invention, and vice versa.

[0075] The specific embodiments described in this invention are examples and do not limit the scope of the invention in any way. For the sake of brevity of the specification, descriptions of prior electronic configurations, control systems, software, and other functional aspects of said systems may be omitted. Additionally, the connections of lines or connecting members between components shown in the drawings are illustrative of functional connections and / or physical or circuit connections, and may be replaced or additionally represented as various functional connections, physical connections, or circuit connections in actual devices. Furthermore, unless specifically stated as “essential,” “importantly,” etc., a component may not be strictly necessary for the application of the invention.

[0076] Furthermore, although the detailed description of the present invention has been explained with reference to preferred embodiments of the invention, those skilled in the art or those with ordinary knowledge in the relevant technical field will understand that various modifications and changes can be made to the invention without departing from the spirit and technical scope of the invention as set forth in the claims below. Accordingly, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification, but should be determined by the claims.

Claims

Claim 1 An opening / closing device installed within a parking area for parking a single vehicle and located adjacent to a speed bump within the parking area; and a server that recognizes a vehicle number within a captured image based on the analysis of a captured image received from the opening / closing device and transmits the result to the opening / closing device; wherein the opening / closing device comprises a main body, a rotating bar connected to a rotating shaft on a rotating bar receiving groove of the main body and rotatable within a predetermined angle range, a controller that controls the rotation of the rotating bar based on the result of recognizing the vehicle number, and a sensing unit for detecting the vehicle and photographing the vehicle, wherein the main body is installed in the parking area, and the rotating bar is installed in the main body but is installed vertically from the main body and is rotatable towards the rear side of the main body around the connection point between the main body and the rotating bar, and the opening / closing device further comprises a main body rotation unit installed on the floor of the parking area to rotate the main body parallel to the ground, and wherein the controller maintains the initial position of the rotating bar when it determines that the vehicle entering the parking area is illegally parked based on the result of recognizing the vehicle number, detects the position of the vehicle in real time, and rotates the main body so that the front of the rotating bar and the rear or front of the vehicle face each other, and wherein the controller [regarding] the vehicle entering the parking area An electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine, configured to maintain the initial position of the rotating bar when the vehicle presses the rotating bar upon determining illegal parking based on the recognition result of the vehicle number. Claim 2 In claim 1, the sensing unit includes an illuminance sensor for measuring illuminance, and the opening / closing device further includes an acoustic output unit for outputting a warning message regarding illegal parking within the parking area, and a light output unit for outputting light when the illuminance value sensed by the illuminance sensor is less than a preset value, thereby utilizing a deep learning-based optical character recognition engine for preventing illegal parking in an electric vehicle charging area. Claim 3 delete Claim 4 An electric vehicle charging area illegal parking prevention system utilizing a deep learning-based optical character recognition engine in claim 1, wherein the controller maintains the initial position of the rotating bar if the result of recognizing the license plate of the vehicle entering the parking area determines that it is illegal parking, and rotates the rotating bar toward the rear side of the main body if it determines that it is normal parking. Claim 5 delete Claim 6 delete Claim 7 A method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine according to claim 1, wherein the opening / closing device detects a vehicle entering the parking area, generates a captured image of the vehicle, and transmits the captured image to a server, the server detects a vehicle license plate area within the captured image, recognizes the vehicle number within the vehicle license plate area, and then looks up the vehicle number, the opening / closing device maintains the initial position of the rotating bar or rotates the rotating bar at a predetermined angle based on the result of looking up the vehicle number, the main body rotating part of the opening / closing device is installed on the floor of the parking area and rotates the main body parallel to the ground, and if the controller of the opening / closing device determines illegal parking based on the result of looking up the vehicle number, it maintains the initial position of the rotating bar, detects the position of the vehicle in real time, and rotates the main body so that the front of the rotating bar and the rear or front of the vehicle face each other, and while the rotating bar maintains the initial position because illegal parking is determined, if the vehicle presses the rotating bar, the rotating bar rotates. Claim 8 A method for preventing illegal parking in an electric vehicle charging area using a deep learning-based optical character recognition engine in claim 7, wherein the server determines the vehicle number within the vehicle number plate area as one of a plurality of types, recognizes the number only up to the number of characters matched to the determined type, and deletes the characters recognized outside the matched number of characters.

Images