KR102531272B1 - Control device using artificial intelligence learning images and vibrations and intrusion alert systme including the same - Google Patents

Abstract

A control device using artificial intelligence includes a database, a sensor interface, a photographing device interface, and a processor, wherein the processor learns sample images using an artificial intelligence algorithm to generate image signal learning data and the image signal learning data stored in the database, generating vibration signal learning data by learning sample vibration signals using the artificial intelligence algorithm, storing the vibration signal learning data in the database, and receiving a first captured image transmitted from a photographing device. and identifies an object included in the first captured image using the image signal learning data, receives a second captured image transmitted from the photographing device, and learns the object included in the second captured image as the image signal. It identifies using data, receives vibration signals output from a sensor that senses vibration through the sensor interface, determines whether the vibration signals are artificial vibration signals using the vibration signal learning data, and determines whether the vibration signals are artificial vibration signals. When the artificial vibration signals are determined, the object is determined as an intruder through the fence, and when the object is determined to be the intruder, the movement of the object identified in each of the first captured image and the second captured image is determined. By analyzing the image signal learning data, a movement method, movement speed, and a movement direction including bearing and angular range of the object are obtained, and the movement direction is analyzed using the image signal learning data so that the object is Predicting which one of a plurality of moving routes will go to the destination, and predicting the time the object will reach the destination by analyzing the moving method and the moving speed using the video signal learning data generating prediction information including the destination and a time to reach the destination, receiving and interpreting the vibration signals from the sensor interface, and generating a control signal for controlling the photographing device according to an analysis result; The photographing device interface receives the control signal for controlling the photographing device from the processor and transmits it to the photographing device.

Description

A control device using artificial intelligence learning images and vibrations and an intrusion alert system including the same

The present invention relates to an intrusion boundary (or intrusion prediction) method using an artificial intelligence algorithm. In particular, a plurality of samples that may occur in a boundary area are learned by using the artificial intelligence algorithm to generate learning data, video signals, A method capable of receiving at least one of sound signals, electrical signals, and vibration signals, analyzing it using the learning data, and predicting the intruder's destination and the time to reach the destination according to the result of the analysis. , a computer program stored in a storage medium for performing the method, and an intrusion detection system for performing the method.

In order to protect the facilities in the main boundary area and the people working or users of the facilities, sensors around the facilities and around the fences that restrict access to the facilities, CCTV cameras operating in conjunction with the sensors , and a boundary system including control devices for controlling the sensors and the CCTV cameras is installed and operated.

This alert system is a reactive response method for intruders, and it is difficult to cope with recently occurring intelligent intrusions.

Registered Patent Publication: Registration No. 10-2021441 (2019.11.04. Notice) Registered Patent Publication: Registration No. 10-1934700 (2019.01.03. Notice) Registered Patent Publication: Registration No. 10-2126498 (2020.06.25. Notice) Registered Patent Publication: Registration No. 10-1951361 (2019.02.22. Notice)

The technical problem to be achieved by the present invention, which was devised to solve a post-reaction method for an intruder, is to generate learning data by learning a plurality of samples using the artificial intelligence algorithm, and to generate video signals, sound signals, and electricity. A method for receiving at least one of signals and vibration signals, analyzing the received signal using the learning data, and predicting how many intruders invade when, where, and with what they possess according to the analysis result, the method described above It is to provide a computer program stored in a storage medium for performing a method, and an intrusion deterrent system for performing the method.

An intrusion detection system according to an embodiment of the present invention includes a photographing device, a sensor for detecting vibration, and a control device using artificial intelligence, wherein the control device using artificial intelligence includes a database and a processor, and the processor Learning sample images using an artificial intelligence algorithm to generate image signal learning data, storing the image signal learning data in the database, learning sample signals using the artificial intelligence algorithm to generate vibration signal learning data, The vibration signal learning data is stored in the database, a first captured image transmitted from the photographing device is received, an object included in the first captured image is identified using the image signal learning data, The transmitted second captured image is received, the object included in the second captured image is identified using the image signal learning data, vibration signals output from the sensor are received, and whether the vibration signals are artificial vibration signals is determined. is determined using the vibration signal learning data, and when the vibration signals are determined to be the artificial vibration signals, the object is determined as an intruder through the fence, and when the object is determined to be the intruder, the first captured image is determined. and analyzing the movement of the object identified in each of the second captured images using the image signal learning data to obtain a movement method, movement speed, and a movement direction including azimuth and angular range of the object, and the movement The direction is analyzed using the video signal learning data to predict whether the object will go to its destination through which one of the plurality of moving paths it will go to, and the moving method and the moving speed are determined by using the video signal learning data. The object is analyzed to predict the time at which the object will arrive at the destination, and generates prediction information including the destination and the time at which the object is to be reached.

According to an embodiment of the present invention, a control device using artificial intelligence includes a database, a sensor interface, a photographing device interface, and a processor, and the processor learns sample images using an artificial intelligence algorithm to generate image signal learning data. generating and storing the image signal learning data in the database, learning sample vibration signals using the artificial intelligence algorithm to generate vibration signal learning data, storing the vibration signal learning data in the database, and transmitting from the photographing device A first captured image is received, an object included in the first captured image is identified using the image signal learning data, a second captured image transmitted from the photographing device is received, and an object included in the first captured image is identified. The object is identified using the image signal learning data, vibration signals output from a sensor for detecting vibration are received through the sensor interface, and whether the vibration signals are artificial vibration signals is determined using the vibration signal learning data. and when the vibration signals are determined to be the artificial vibration signals, the object is determined to be an intruder through the fence, and when the object is determined to be the intruder, in each of the first captured image and the second captured image. The movement of the identified object is analyzed using the video signal learning data to obtain a movement method, movement speed, and a movement direction including bearing and angular range of the object, and the movement direction is determined by using the video signal learning data. It predicts whether the object will go to the destination through which one of the plurality of movement routes it will go to, and analyzes the movement method and the movement speed using the video signal learning data so that the object can predicting a time to reach a destination, generating prediction information including the destination and a time to reach the destination, receiving and interpreting the vibration signals from the sensor interface, and controlling the photographing device according to the analysis result and the photographing device interface receives the control signal for controlling the photographing device from the processor and transmits the control signal to the photographing device.

A method according to an embodiment of the present invention, a computer program stored in a storage medium for performing the method, and an intrusion alert system for performing the method according to an embodiment of the present invention devised to solve a post-reaction method for an intruder, Learning using an intelligent algorithm to generate learning data, receiving at least one of images, sound signals, electrical signals, and vibration signals, analyzing the received signal using the learning data, and It has the effect of predicting how many intruders will invade when, where, and with what they possess.

Accordingly, the method, the computer program, and the intrusion detection system prevent an intruder from entering a fence or facility, arrive at an intrusion route or an escape route of the intruder in advance, and take quick action against the intruder. there is

A detailed description of each drawing is provided in order to more fully understand the drawings cited in the detailed description of the present invention.
1 is a block diagram of an intrusion detection system using artificial intelligence according to an embodiment of the present invention.
FIG. 2 is a block diagram of a control device using artificial intelligence shown in FIG. 1 .
FIG. 3 is a flowchart illustrating an embodiment of an operating method of the intrusion detection system shown in FIG. 1 .
4 is a diagram illustrating a process of generating learning data using an artificial intelligence algorithm.
5 is a diagram for explaining a method of acquiring a moving method, a moving speed, and a moving direction of an object using captured images.
6 is a flowchart illustrating the operation of an intrusion deterrent system when an object is determined to be an intruder.
FIG. 7 is a flowchart illustrating an embodiment of an operating method of the intrusion detection system shown in FIG. 1 .

1 is a block diagram of an intrusion detection system using artificial intelligence according to an embodiment of the present invention.

Referring to FIG. 1, an intrusion route and/or an escape route of an intruder 500 to a fence 400 or facility 410 is predicted in real time using artificial intelligence (AI), and the result of the prediction An intrusion boundary capable of generating predictive information (IF) according to (this is also referred to as 'intrusion monitoring' or 'intrusion prediction'). The system 100 is a control device using artificial intelligence (or an artificial intelligence program) (this ' 200), a plurality of sensors (311, 321, 331, 341, 351, 361, 371, 381, and 391), a plurality of photographing devices (313, 323, 333, 343, 353, 363, 373, 383, and 393), a fence 400, a facility 410, a first object 500, and a terminal 600, and according to embodiments, the intrusion boundary system 100 includes a camera May further include a drone 700 equipped with.

The intrusion detection system 100 includes (i) an input signal related to the behavior of the first object 500 (eg, a video signal IM, a sound signal SS, an electrical signal ES, and a vibration signal VS). ), or (ii) analyzing the input signal and feature information of the area where the first object 500 moves using the learning data 241, and according to the result of the analysis, the first object 500 ) is from the first position P1 when (eg, year, month, hour, minute, second), where (eg, a certain part (DT1, DT2, or DT3) of the fence 400, or a certain part of the space inside the boundary ( For example, DT4 or DT5)), and through any path (eg, PA1, PA2, PA3, PB1, PB2, and PC1)) trespassing into the fence 400 or facility 410, and then when (e.g. year month date hour minute second) where (e.g. which part of fence 400 (DT6 or DT7)), and by which route (e.g. PD1, PD2, and PE1) escapes It is a system that predicts what to do in real time and notifies the result of the prediction (IF) to the terminal 600 to prevent intrusion of the first object 500 or to take quick action on the first object 500 .

The artificial intelligence control device 200 uses a neural network (Neural Networks) algorithm, a machine learning (Machine Learning) algorithm, or a deep learning (Deep Learning) algorithm, etc. , sample sound signals, sample electrical signals, and sample vibration signals) by learning (eg, multi-instance learning) to obtain learning data 241, which is referred to as 'big data'. )') is generated, and the learning data 241 is stored in the database 240.

According to an embodiment, various samples for learning by the artificial intelligence program 212 may be data labeled data. Data labeling refers to the task of attaching names (or labels) to samples (or source data) so that artificial intelligence can learn a lot of unstructured data.

The artificial intelligence control device 200 receives a detection signal from at least one of the sensors 311, 321, 331, 341, 351, 361, 371, 381, and 391 (at least one of SS, ES, and VS, which is 'SS' , ES, and/or VS') is received and analyzed, and according to the result of the analysis, at least one of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 operation can be controlled.

Each of the sensors 311 , 321 , 331 , 341 , 351 , 361 , 371 , 381 , and 391 is a sensor for detecting an intrusion or movement of the first object 500 . The type of each sensor 311, 321, 331, 341, 351, 361, 371, 381, and 391 is a sensor that detects infrared rays, a sensor that detects laser, a sensor that detects radio waves, and a sensor that detects heat. sensor, motion sensor, light sensor, pressure sensor, sound sensor, electrical signal sensor, vibration sensor, gas sensor, weather sensor A sensor, and an object image sensor, and the like. Speakers or sirens may be installed around each of the sensors 311 , 321 , 331 , 341 , 351 , 361 , 371 , 381 , and 391 .

In this specification, each of the sensors 311, 321, 331, 341, 351, 361, 371, 381, and 391 collectively refers to a sensor including at least one of the sensors performing the functions listed above.

At least one of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 is a CCTV camera, a network camera, or a pan-tilt-zoom (PTZ) camera. , a thermal imaging camera, or a thermal observation device, but is not limited thereto. A thermal imaging camera is a device that generates an image using infrared radiation, and is also called an infrared camera, a thermal imaging camera, or a thermal imager.

The paired sensors and photographing devices 311 and 313 , 321 and 323 , to 391 and 393 may operate in conjunction with each other.

The facility 410 is surrounded by a fence 400 and borders (or Expenses) means facilities that require

Each movement path (PA1, PA2, PA3, PB1, PB2, PB3, and PC1) for access to the facility 410 may be referred to as an intrusion route, and each movement route (PD1, PD2) for escape from the facility 410 , and PE1) can be referred to as escape routes.

Although one fence 400 surrounding the facility 410 is shown in FIG. 1 , the number of fences 400 surrounding the facility 410 may be two or more according to embodiments.

The first object 500 may be a moving person or a moving wild animal, but in this specification, it is assumed that the first object 500 is a person (or an intruder).

The terminal 600 refers to a communication device capable of receiving prediction information (IF, eg, information including a predicted destination and a predicted arrival time) from the artificial intelligence control device 200 .

Examples of the terminal 600 include a communication device possessed by a person in charge of guarding (or guarding) the fence 400 and the facility 410 (eg, security guards, soldiers, police officers, and/or police); A communication device of a border company, a communication device of a military base, a communication device of a police station, or a communication device of a fire department, but is not limited thereto.

When the terminal 600 receives the prediction information (IF) from the artificial intelligence control device 200, the person in charge of guarding the fence 400 uses the prediction information (IF) to locate the first object 500 (DT1). At least one of ~ DT8) has an effect of preventing intrusion of the first object 500 or taking quick measures (eg, arrest, etc.) against intrusion of the first object 500 .

The drone 700 equipped with a camera captures an image of the first object 500 in real time under the control of the artificial intelligence control device 200 or the manager of the artificial intelligence control device 200, and displays the captured images. It can be transmitted in real time through wireless communication to the artificial intelligence control device 200.

FIG. 2 is a block diagram of a control device using artificial intelligence shown in FIG. 1 .

Referring to FIGS. 1 and 2 , the artificial intelligence control device 200 includes an input device 205 and a processor 210 that executes an artificial intelligence program 212 that operates according to an artificial intelligence algorithm 210, which is also referred to as a 'server'. ), a memory device 215, a sensor interface 220, a photographing device interface 230, a database 240, a communication device 250, a display device 270, and an alarm device 280, Accordingly, the artificial intelligence control device 200 may further include a drone control device 260.

The artificial intelligence program 212 combined with hardware to perform the intrusion prediction method using the artificial intelligence algorithm described herein is stored in a storage medium accessible by the processor 210 (e.g., memory device 215). and is executed by the processor 210.

The sensor interface 220 receives detection signals SS, ES, and/or VS output from the corresponding sensors 311, 321, 331, 341, 351, 361, 371, 381, and 391, and receives the detection signal ( SS, ES, and/or VS) to the artificial intelligence program 212.

The artificial intelligence program 212 receives and analyzes the detection signals (SS, ES, and/or VS), and according to the interpretation result, the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393) to generate a control signal (CTL) for controlling at least one photographing device (for example, when the corresponding photographing device is a PTZ camera, controlling at least one of pan, tilt, and zoom of the PTZ camera); , The control signal CTL is transmitted to the at least one photographing device through the photographing device interface 230 .

The artificial intelligence program 212 receives the captured images from the at least one photographing device through the photographing device interface 230, and performs the steps (or operations) to be described with reference to FIGS. 3, 5, 6, and 7 s) are performed.

For example, when the first object 500 is detected by the sensor 311 installed at the first location P1, the artificial intelligence program 212 outputs the detection signals SS, ES, and A control signal CTR is generated in response to / or VS, and the control signal CTR is transmitted to the photographing device 313 installed at the first position P1 through the photographing device interface 230.

When the photographing device 313 is a PTZ camera, the photographing device 313 controls at least one of pan, tilt, and zoom to capture the first object 500 in response to the control signal CTR, and The photographed images of 500 are transmitted to the artificial intelligence program 212 through the photographing device interface 230 .

In addition, when the first object 500 is detected by the corresponding sensors 321, 331, 341, 351, 361, 371, 381, and 391, the artificial intelligence program 212 sends the corresponding photographing device 323, 333, 343 , 353, 363, 373, 383, and 393) to generate a control signal (CTR), and transmit the control signal (CTR) to the corresponding photographing devices (323, 333, 343, 353, 363, 373, 383, and 393).

According to embodiments, the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 may have characteristics of the first object 500 (eg, the head of the first object 500). , face, or feet) may be tracked in real time to generate captured images.

In addition, when the first object 500 is detected by the sensor 311 installed at the first location P1, the artificial intelligence program 212 outputs the detection signal (SS, ES, and/or When a control signal CTR capable of controlling each of the two photographing devices 313 and 343 is generated in response to VS, the control signal CTR may be transmitted to the photographing devices 313 and 343. Accordingly, both of the photographing devices 313 and 343 may focus (or photograph) the third position P3 where the first object 500 exists.

The database 240 is the location (or region) where the learning data 241 to be described with reference to FIGS. 3 and 4 and each of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 are installed. The feature information 243 for ) may be stored. According to embodiments, the database 240 may refer to a data storage device that stores information 241 and 243 or information stored in the data storage device.

As shown in FIG. 2, feature information 243 is sensor information 311I, 321I for sensors 311, 321, 331, 341, 351, 361, 371, 381, and 391 for each ID (ID1 to ID9). , 331I, 341I, 351I, 361I, 371I, 381I, and 391I, and imaging device information (313I, 323I, and 333I for the imaging devices 313, 323, 333, 343, 353, 363, 373, 383, and 393) , 343I, 353I, 363I, 373I, 383I, and 393I), location information (PI1 to PI9), terrain information (TG1 to TG9), and feature information (FI1 to FI9).

Each sensor information (311I, 321I, 331I, 341I, 351I, 361I, 371I, 381I, and 391I) is information on the type of the corresponding sensor, and each photographing device information (313I, 323I, 333I, 343I, 353I, 363I, 373I, 383I, and 393I) is information about the type of the corresponding photographing device, and each position information (PI1 to PI9) is information about the photographing device 313, 323, 333, 343, 353, 363, 373, 383, and 393 is installed location information, and may be, for example, global positioning system (GPS) coordinates.

Since topography refers to the shape or shape of the land, the topography information (TG1 to TG9) includes information on whether the land is flat, mountainous, sloped, wetland, or an area with oddly shaped cliffs.

Since features refer to all natural or artificial objects existing on the ground, feature information (FI1 to FI9) includes information on trees, rocks, houses, rivers, valleys, and roads.

According to embodiments, the artificial intelligence program 212 may determine from which of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 the captured images are received. , When predicting the destination of the first object 500 and the time to reach the destination, the feature information 243 can be utilized.

According to embodiments, the artificial intelligence program 212 may determine from which of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 the captured images are received. , At least one of the feature learning data 241E and the feature information 243 may be used when predicting the destination of the first object 500 and the time to reach the destination.

According to embodiments, the artificial intelligence program 212 learns the terrain and/or features included in each of the sample images (SPI) using an artificial intelligence algorithm, and generates feature learning data 241E according to the learning result. And, the feature learning data 241E may be stored in the database 240 .

The communication device 250 is a device for communication with the terminal 600 or the drone 700.

Under the control of the artificial intelligence program 212, the display device (or monitor, 270) is transmitted from at least one of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393. The captured images may be received and displayed.

For example, the photographing device interface 230 communicates with each of the photographing devices 313, 323, 333, 343, 353, 363, 373, 383, and 393 through different communication channels, and communicates with the photographing device 313. When the first object 500 is detected by the paired sensors 311, the artificial intelligence program 212 takes pictures based on the detection signals SS, ES, and/or VS output from the sensor 311. A communication channel between the device interface 230 and the photographing device 313 is automatically connected.

Accordingly, the display device 270 may receive captured images output from the photographing device 313 and display them on the display device 270 under the control of the artificial intelligence program 212 .

When the artificial intelligence program 212 determines that the first object 500 is an intruder, the artificial intelligence program 212 outputs an alarm signal to the alarm device 280 . Alert device 280 may be an audible alert device, a visual alert device, or an audible alert device.

The drone control device 260 may control the operation of the drone 700 . When the artificial intelligence program 212 determines that the first object 500 is an intruder, the artificial intelligence program 212 generates a drone control signal and outputs it to the drone control device 260, the drone control signal communicates with It is transmitted to the drone 700 through the device 250 .

When the artificial intelligence program 212 generates a drone control signal based on the detection signal (SS, ES, and/or VS) output from the sensor 311 or the captured image (IM) output from the photographing device 313 , The drone control signal includes the location information PI1 of the sensor 311 or the location information PI1 of the photographing device 313.

Therefore, the drone 700 moves to a location corresponding to the location information PI1 of the sensor 311 or the location information PI1 of the photographing device 313 in response to the drone control signal transmitted from the artificial intelligence control device 200. It flies automatically and captures an image of the first object 500 while tracking the first object 500 and transmits the captured image to the communication device 250 . The artificial intelligence program 212 receives the captured image of the first object 500 captured by the drone 700 through the communication device 250 and transmits it to the display device 270 .

According to another embodiment, the artificial intelligence program 212 controls the detection signal SS, ES, and/or VS output from the sensor 311 or the photographed image IM output from the photographing device 313. 1 When determining the object 500 as an intruder, the manager of the artificial intelligence control device 200 manipulates the drone control device 260 to send the drone 700 to the location information (PI1) of the sensor 311 or the shooting device ( 313), the first object 500 can be tracked in real time by flying to a location corresponding to the location information PI1.

The captured image of the first object 500 photographed by the drone 700 is transmitted to the artificial intelligence program 212 through the communication device 250, and the artificial intelligence program 212 is photographed by the drone 700. A photographed image of the first object 500 is received and transmitted to the display device 270 .

3 is a flowchart illustrating an embodiment of an operating method of the intrusion detection system shown in FIG. 1, and FIG. 4 is a diagram illustrating a process of generating learning data using an artificial intelligence algorithm.

1 to 4, the artificial intelligence program 212 includes sample images (SPI), sample sound signals (SDS), sample electrical signals (ELS), and sample vibrations, each of which includes at least one object. The signals (VBS) or a combination of at least two of these (SPI, SDS, ELS, and VBS) are learned using an artificial intelligence algorithm, and learning data 241 is generated according to the learning result and stored in the database 240. Save (S110).

A process of generating the learning data 241 by the artificial intelligence program 212 will be illustratively described with reference to FIG. 4 .

The learning data 241 includes image signal learning data 241A generated according to the learning result of the sample images SPI, sound signal learning data 241B generated according to the learning result of the sample sound signals SDS, and sample It includes at least one of electrical signal learning data 241C generated according to the learning result of the electrical signals ELS and vibration signal learning data 241D generated according to the learning result of the sample vibration signals VBS. According to an embodiment, the learning data 241 may further include feature learning data 241E.

As shown in FIG. 4, a plurality of samples for learning (eg, sample images (SPI), sample sound signals (SDS), sample electrical signals (ELS), and sample vibration signals (VBS)) at least one of them) is input into the artificial intelligence program 212 through the input device 205. According to embodiments, samples for learning received through the communication device 250 are input to the artificial intelligence program 212 .

The artificial intelligence program 212 sets an object (or a plurality of objects) included in each image signal (IMS) corresponding to each sample image (SPI) to a human element (HME), an animal element (ANE), and a natural environment element ( NEE), artificial component (ATE), and noise component (NSE) are trained (or sorted). The elements (HME, ANE, NEE, ATE, and NSE) shown in FIG. 4 are only illustrated for explanation.

For example, the artificial intelligence program 212 learns (or sorts) at least one object included in each sample image SPI corresponding to each sample image signal IMS into the following elements.

1. When an object included in each sample image (SPI) is a human, the artificial intelligence program 212 learns the object as a human element (HME);

2. When the object is an animal, AI program 212 learns the object as an animal element (ANE);

3. When the object is lightning, artificial intelligence program 212 learns the object as a natural environment element (NEE);

4. When the object is a light or a spark generated when cutting the wire of the fence 400 with a tool (eg, cutter), the artificial intelligence program 212 learns the object as an artificial element (ATE); and

5. When the object is a forest fire, the artificial intelligence program 212 learns the object as a noise component (NSE).

In addition, the artificial intelligence program 212 converts each of the sample sound signals SDS into a human component (HME), an animal component (ANE), a natural environment component (NEE), an artificial component (ATE), and a noise component (NSE). Learn (or sort).

For example, the artificial intelligence program 212 learns (or sorts) each sample sound signal (SDS) into the following elements.

1. The artificial intelligence program 212 learns each sample sound signal (SDS) corresponding to the sound of human footsteps as a human element (HME);

2. The artificial intelligence program 212 learns each sample sound signal (SDS) corresponding to the sound of animal footsteps as an animal element (ANE);

3. The artificial intelligence program 212 learns each sample sound signal (SDS) corresponding to the sound of wind, rain, waves, the sound of a stone rolling in waves, or the sound of flowing water as a natural environment element (NEE);

4. The artificial intelligence program 212 generates a gunshot sound, a moving sound of a vehicle, a sound of digging the ground, a sound of a person climbing the fence 400, or a sound of cutting the wire of the fence 400 with a cutter. Learning each sample sound signal (SDS) corresponding to a sound as an artifact (ATE);

5. The artificial intelligence program 212 learns each sample sound signal (SDS) corresponding to thunder as a noise component (NSE).

In addition, the artificial intelligence program 212 learns (or sorts) each of the sample electrical signals ELS into a natural environment component NEE, an artificial component ATE, and a noise component NSE.

For example, the artificial intelligence program 212 learns (or sorts) each sample electrical signal (ELS) into the following elements.

1. The artificial intelligence program 212 learns each sample electrical signal (ELS) generated by lightning as a natural environment element (NEE);

2. The artificial intelligence program 212 generates each electrical signal (ELS) applied to cut the fence 400 and each sample electrical signal (ELS) applied to neutralize the electric line or communication line installed around the fence 400 learning with artificial elements (ATE);

3. The artificial intelligence program 212 learns each sample electrical signal (ELS) generated by static electricity as a noise component (NSE).

In addition, the artificial intelligence program 212 converts each of the sample vibration signals (VBS) into a human component (HME), an animal component (ANE), a natural environment component (NEE), an artificial component (ATE), and a noise component (NSE). Learn (or sort).

For example, the artificial intelligence program 212 learns (or sorts) each sample vibration signal (VBS) into the following elements.

1. The artificial intelligence program 212 learns each sample vibration signal (VBS) generated by human motion as a human element (HME);

2. The artificial intelligence program 212 learns each sample vibration signal (VBS) generated by the animal's motion as an animal element (ANE);

3. The artificial intelligence program 212 learns each sampled vibration signal (VBS) generated by a lightning strike, rolling cloud, or falling tree as a natural environment element (NEE);

4. The artificial intelligence program 212 learns each sample vibration signal (VBS) generated by demolition, vibration generated when digging with a tool, and movement of a vehicle as an artificial factor (ATE);

5. The artificial intelligence program 212 learns each sample vibration signal (VBS) generated by the seismic wave as a noise component (NSE).

The artificial intelligence program 212 is an object (eg For example, at least one of the movement method (MM), movement speed (MV), movement direction (MD), and intrusion type (IT) of a person or animal) is learned, and the object exists using the samples. Learn local landmarks (LM).

For example, the artificial intelligence program 212 learns an object movement method (MM) using learned samples. For example, the artificial intelligence program 212 learns the movement method (MM) of the object by learning the movement of the same object included in each of the sample images (IMS). Here, examples of the movement method (MM) include whether the object moves by crawling, whether the object moves by walking, whether the object moves by jumping, or whether the object moves by means of movement (eg, a vehicle or a boat, etc.) , and whether the object moves by swimming.

The artificial intelligence program 212 learns the moving speed (MV) of the object using the learned samples. For example, the artificial intelligence program 212 learns the movement speed (MV) of the object by learning the movement of the same object included in each of the sample images (IMS).

The artificial intelligence program 212 learns the movement direction (MV) of the object using the learned samples. For example, the artificial intelligence program 212 learns the movement direction (MV) of the object by learning the movement direction of the same object included in each of the sample images (IMS).

The artificial intelligence program 212 uses learned samples (eg, learning results for the samples (IMS, SDS, ELS, or VBS), and a combination of at least two of the learning results) Learn intrusion types (IT).

For example, the artificial intelligence program 212 learns the intrusion type (IT) of the object according to the movement of the same object included in each of the sample images (IMS).

Examples of the intrusion type (IT) include whether the object climbs the fence 400, jumps over the fence 400, climbs the fence 400 using a ladder, and It learns whether to dig a hole under the burrow, knock over the fence 400, or cut a hole through the fence 400.

The artificial intelligence program 212 may combine the following learning results in order to increase the accuracy of determining the intrusion type (IT). The more learning results to combine, the higher the accuracy of the judgment on the intrusion type (IT).

1. Combination of learning results for sample images (IMS) and learning results for sample sound signals (SDS);

2. Combination of learning results for the sample images IMS and learning results for the sample electrical signals ELS;

3. Combination of learning results for the sample images (IMS) and learning results for the sample vibration signals (VBS);

4. A combination of the learning result of the sample images IMS, the learning result of the sample sound signals SDS, and the learning result of the sample electrical signals ELS;

5. A combination of a learning result for sample images (IMS), a learning result for sample sound signals (SDS), and a learning result for sample vibration signals (VBS);

6. The learning results for sample images (IMS), the learning results for sample sound signals (SDS), the learning results for sample electrical signals (ELS), and the learning results for sample vibration signals (VBS) Combination.

Of course, in preparation for failure or intentional failure of the image capture device, the artificial intelligence program 212 may learn the intrusion type (IT) of the object using the following examples.

(1) learning results for sample sound signals (SDS);

(2) learning results on sample electrical signals (ELS);

(3) learning results for the sample vibration signals (VBS);

(4) A combination of at least two of (1) to (3).

The artificial intelligence program 212 uses learned samples (eg, learning results for the samples (IMS, SDS, ELS, or VBS), and a combination of at least two of the learning results) learn (LM).

For example, the artificial intelligence program 212 learns the feature LM according to objects included in each of the sample images IMS. For example, the artificial intelligence program 212 may learn the feature LM based on the distribution of trees included in each of the sample images IMS, whether there is a valley, whether water flows, whether there is a slope, and the like. there is.

5 is a diagram for explaining a method of acquiring a moving method, a moving speed, and a moving direction of an object using captured images.

The first object 500 indicated by a dotted line in the corresponding captured images FA2, FB2, and FC2 means the first object 500 included in the previous images FA1, FB1, and FC1, and the first object 500 indicated by a solid line The object 500 refers to the first object 500 included in the corresponding captured images FA2 , FB2 , and FC2 .

In order to facilitate understanding of the present invention, the currently captured images FA2, FB2, and FC2 include the first object 500 indicated by dotted lines included in the previously captured images FA1, FB1, and FC1, and the currently captured image ( The first object 500 indicated by a solid line included in FA2, FB2, and FC2) is shown together.

Referring to case 1 (CASE1) according to an embodiment, after learning data 241 is generated and stored in the database 240, the artificial intelligence program 212 receives the first captured image FA1 at a first point in time and The first object 500 included in the first captured image FA1 is identified using the learning data 241 (S115).

After that, the artificial intelligence program 212 receives the second captured image FA2 at the second time point and identifies the first object 500 included in the second captured image FA2 using the learning data 241. (S120).

For example, when performing each step (S115 and S120), the artificial intelligence program 212 determines how many objects are included in each of the captured images (FA1 and FA2), and determines whether the object is a tool (eg, a sword or a gun). , a cutter, or a ladder) may be identified using the learning data 241 .

The artificial intelligence program 212 analyzes the movement (or change amount of the movement) of the first objects 500 identified in steps S115 and S120 using the learning data 241, and the first object 500 A first moving method, a first moving speed, and a first moving direction of are obtained (S125).

At this time, the artificial intelligence program 212 uses a specific part of the first object 500 (eg, the head, foot, or hand of the first object 500) to move the first objects 500 ( Alternatively, the amount of change in motion) may be analyzed using the learning data 241 .

The artificial intelligence program 212 predicts the first destination DT1 of the first object 500 by analyzing the first moving direction (for example, north-north-east, several degrees to several degrees) using the learning data 241. do. For example, the artificial intelligence program 212 predicts that the first object 500 will go to the first destination DT1 through the first movement path PA1.

The artificial intelligence program 212 analyzes the first moving method (for example, trotting) and the first moving speed (70 to 80 meters per minute) using the learning data 241 so that the first object 500 is A first time to arrive at one destination (DT1) is predicted (S135). The predicted first time includes a minimum predicted time and a maximum predicted time for the first object 500 to reach the first destination DT1 from the first location P1 through the first movement path PA1.

For example, referring to case 2 (CASE2), the artificial intelligence program 212 receives the first captured image FB1 at a first point in time and displays the first object 500 included in the first captured image FB1. It is identified using the learning data 241 (S115).

Then, the artificial intelligence program 212 receives the second captured image FB2 at a second time point and identifies the first object 500 included in the second captured image FB2 using the learning data 241. (S120).

The artificial intelligence program 212 analyzes the movement (or the amount of change in movement) of the objects 500 identified in steps S115 and S120 using the learning data 241, and the first object 500 The second movement method, the second movement speed, and the second movement direction are acquired (S125).

The artificial intelligence program 212 predicts the second destination DT2 of the first object 500 by analyzing the second moving direction (eg, west-northwest, several degrees to several degrees) using the learning data 241. do. For example, the artificial intelligence program 212 predicts that the first object 500 will go to the second destination DT2 through the second movement path PA2 including the second point P2.

The artificial intelligence program 212 analyzes the second moving method (eg, urgent report) and the second moving speed (eg, 100 to 110 meters per minute) using the learning data 241 to obtain the first object ( 500) predicts a second time to reach the second destination DT2 via the second point P2 (S135). The predicted second time includes a minimum predicted time and a maximum predicted time for the first object 500 to reach the second destination DT2 from the first location P1 through the movement paths PA2 and PB2.

For example, referring to case 3 (CASE3), the artificial intelligence program 212 receives the first captured image FC1 at a first point in time and displays the first object 500 included in the first captured image FC1. It is identified using the learning data 241 (S115).

Then, the artificial intelligence program 212 receives the second captured image FC2 at a second time point and identifies the first object 500 included in the second captured image FC2 using the learning data 241. (S120).

The artificial intelligence program 212 analyzes the movement (or the amount of change in movement) of the objects 500 identified in steps S115 and S120 using the learning data 241, and the control of the first object 500 3 The movement method, the third movement speed, and the third movement direction are acquired (S125).

The artificial intelligence program 212 predicts the third destination DT3 of the first object 500 by analyzing the third moving direction (eg, east-north-east, several degrees to several degrees) using the learning data 241. do. For example, the artificial intelligence program 212 predicts that the first object 500 will go to the third destination DT3 through the third movement path PA3 including the third point P3.

The artificial intelligence program 212 analyzes the third moving method (eg, jumping) and the third moving speed (eg, 15 km to 18 km per hour) using the learning data 241 to obtain the first object 500 predicts the third time to reach the third destination DT3 via the third point P3 (S135). The predicted third time includes a minimum predicted time and a maximum predicted time for the first object 500 to reach the third destination DT3 from the first location P1 through the movement paths PA3 and PB3.

Fig. 6 is a flowchart describing the operation when an object is determined to be an intruder. 1 to 6, cases in which the artificial intelligence program 212 determines the first object 500 that has reached the third destination DT3 as an intruder will be described.

The sensor 341 may detect at least two of (i) a sensor for detecting sound, (ii) a sensor for detecting an electrical signal, (iii) a sensor for detecting vibration, or (iv) a sensor for detecting sound, an electrical signal, and vibration. collectively refers to a set of sensors that

Examples in which the artificial intelligence program 212 can determine the first object 500 as an intruder are as follows.

(1) Determining an intruder using the images IM captured by the photographing device 343;

(2) Determining an intruder by using the sound signals SS output from the sensor 341 for detecting sound;

(3) determining an intruder using the electrical signals ES output from the sensor 341 that detects electrical signals;

(4) determining an intruder using the vibration signals VS output from the sensor 341 that senses vibration;

(5) Determined as an intruder by combining at least two of (1) to (4).

For example, the artificial intelligence program 212 receives each of the third captured images captured by the photographing device 343 through the photographing device interface 230, and receives the first captured images included in each of the third captured images. The object 500 and the second object are identified using the learning data 241 (S115 and S120), and the movement of the first object 500 included in each of the third captured images is measured using the learning data 241. The fourth movement method, the fourth movement speed, and the fourth movement direction of the first object 500 are obtained (S125), and the fourth movement direction of the first object 500 is obtained (S125). The destination DT4 is predicted (S130), and the fourth time when the first object 500 will arrive at the fourth destination DT4 is predicted using the fourth moving method and the fourth moving speed (S135). . The predicted fourth time includes a minimum predicted time and a maximum predicted time for the first object 500 to reach the fourth destination DT4 from the third destination DT3.

When the second object is identified as the fence 400 by the artificial intelligence program 212, the artificial intelligence program 212 determines whether the first object 500 is intruding through the fence 400 (S137). .

When it is determined that the first object 500 is a person (ie, an intruder) and that the intruder 500 is trespassing on the fence 400 (YES in S137), the artificial intelligence program 212 generates an alarm device 280 ) is used to generate an alarm, and generate prediction information (IF) including the fourth destination (DT4) and the fourth time for the first object 500 to arrive at the fourth destination (DT4) so that the terminal 600 send to The terminal 600 may generate an alarm (eg, a visual alarm, an audible alarm, or an audio-visual alarm) using an alarm of the terminal 600 in response to the prediction information IF.

According to an embodiment, the artificial intelligence program 212 identifies and analyzes the third captured images captured by the photographing device 343 using the learning data 241, and according to the analysis result, the first object 500 When it is possible to determine how many people, whether the first object 500 possesses a tool (for example, a sword, gun, cutter, ladder, etc.), and the type of the tool, the artificial intelligence program 212 performs a fourth Destination DT4, 4th time, number of first objects 500, whether first object 500 has a tool, and prediction information IF including the type of the tool is generated, and the terminal 600 ) is sent to

Accordingly, the person in charge of the terminal 600 can prevent the intrusion of the first object 500 as an intruder or establish measures (or countermeasures) for the intrusion of the first object 500 according to the prediction information IF. . Actions (or countermeasures) include how many officers are needed and what weapons each of those officers will carry.

According to an embodiment, the artificial intelligence program 212 identifies and analyzes the third captured images captured by the photographing device 343 using the learning data 241, and according to the analysis result, the first object 500 When it is possible to determine how many people, whether the first object 500 possesses a tool (eg, a sword, a gun, or a cutter), and the type of the tool, the artificial intelligence program 212 determines the fourth destination ( DT4), 4th time, how many people the first object 500 has, whether the first object 500 has a tool, the type of the tool, and preventing intrusion by the first object 500 or the first object ( 500), predictive information (IF) including how many officers are needed and which weapons each officer will carry is generated and transmitted to the terminal 600.

That is, the artificial intelligence program 212 analyzes measures (or countermeasures) to respond to the intruder, the first object 500, using the learning data 241, and takes the measures (or countermeasures) according to the analysis result. Prediction information (IF) including may be generated and transmitted to the terminal 600.

Examples of trespassing by the first object 500 include unauthorized cutting of the fence 400 by the first object 500, passing through a hole in the fence 400 by the first object 500, The act of the first object 500 digging a burrow under the fence 400, the first object 500 climbing the fence 400 on a ladder, and the first object 500 crossing the fence 400 includes action

The artificial intelligence program 212 analyzes the behavior (or behavior) of the first object 500 included in each of the third captured images using the learning data 241, and according to the result of the analysis, the first object 500 ) is a person, but when it is determined that the person does not invade the fence 400 (NO in S137), the artificial intelligence program 212 is a speaker installed around the sensor 341 through the sensor interface 220. It is possible to broadcast a warning by controlling (S139).

According to an embodiment, when it is determined that the first object is an animal and the animal does not invade the fence 400 (NO in S137), the artificial intelligence program 212 operates the sensor 341 through the sensor interface 220. ) can be warned by controlling the warning lights installed around it. According to another embodiment, when the first object is an animal and the fence 400 is an electric fence (NO in S137), the artificial intelligence program 212 causes current to flow through the sensor interface 220 to the electric fence 400 to The animal can be expelled from the electric fence 400.

According to embodiments, when the second object is identified as the fence 400, the artificial intelligence program 212 determines that the first object 500 is intruding through the fence 400 using sound signals SS It does (S137).

Since the learning data 241 includes the sound signal learning data 241B, the artificial intelligence program 212 receives the sound signals SS output from the sensor 341 through the sensor interface 220, and the sound signal It is determined whether the sound signals SS are artificial sound signals using the sound signal learning data 241B, and when the sound signals SS are determined to be the artificial sound signals, the first object 500 is moved to the fence 400. Determined as an intruder through (YES in S137), generates an alarm using the alarm device 280, and the fourth destination (DT4) and the first object 500 reach the fourth destination (DT4). Prediction information (IF) including time is generated and transmitted to the terminal 600 (S140).

According to embodiments, when the second object is identified as the fence 400, the artificial intelligence program 212 determines that the first object 500 is intruding through the fence 400 using electrical signals ES. It does (S137).

Since the learning data 241 includes the electrical signal learning data 241C, the artificial intelligence program 212 receives the electrical signals ES output from the sensor 341 through the sensor interface 220, and receives the electrical signal It is determined whether the ESs are artificial electrical signals using the electrical signal learning data 241C, and when the electrical signals ES are determined to be artificial electrical signals, the first object 500 is transferred to the fence 400. Determined as an intruder through (YES in S137), generates an alarm using the alarm device 280, and the fourth destination (DT4) and the first object 500 reach the fourth destination (DT4) Prediction information (IF) including time is generated and transmitted to the terminal 600 (S140).

According to embodiments, when the second object is identified as the fence 400, the artificial intelligence program 212 determines whether the first object 500 is intruding through the fence 400 using the vibration signals VS It does (S137).

Since the learning data 241 includes the vibration signal learning data 241D, the artificial intelligence program 212 receives the vibration signals VS output from the sensor 341 through the sensor interface 220, and the vibration signal It is determined whether VS is artificial vibration signals using the vibration signal learning data 241D, and when the vibration signals VS are determined to be the artificial vibration signals, the first object 500 is moved to the fence 400. Determined as an intruder through (YES in S137), generates an alarm using the alarm device 280, and the fourth destination (DT4) and the first object 500 reach the fourth destination (DT4). A prediction signal (IF) including time is generated and transmitted to the terminal 600 (S140).

As described in the previous embodiments, after the destination and the time to reach the destination are predicted, the artificial intelligence program 212 captures images captured by the corresponding photographing device 353, 363, 373, 383, or 393. The first object 500 included in each is identified (S145), and the movement of the first object 500 is analyzed to predict a new destination of the first object 500 and a new time to reach the new destination, The predicted destination and the predicted time are updated in real time with the predicted new destination and the predicted new time (S150), and new prediction information (IF) including the predicted new destination and the predicted new time is generated to communicate with the communication device. It is transmitted to the terminal 600 through (250) (S140).

The artificial intelligence program 212 predicts the destination (DT5, DT6, DT7, or DT8) and the time to reach the destination (DT5, DT6, DT7, or DT8) of the first object 500, and the predicted destination and the predicted destination The process of generating the prediction information (IF) including time includes the destination (DT1, DT2, DT3, or DT4) of the first object 500 and the time to reach the destination (DT1, DT2, DT3, or DT4). It is almost the same as the process of predicting and generating prediction information (IF) including the predicted destination and the predicted time.

FIG. 7 is a flowchart illustrating an operation method of the intrusion detection system shown in FIG. 1 .

Referring to FIGS. 1 to 7 , the artificial intelligence program 212 identifies a terrain and/or feature for the location where the first object 500 exists, and uses the identified terrain feature to reach a destination and the destination. Describe how to predict time. It is assumed that the object 510 included in each of the captured images FA1 , FA2 , FB1 , FB2 , FC1 , and FC2 of FIG. 5 is an object for acquiring information on a terrain feature.

According to embodiments, referring to case 1 (CASE1) of FIG. 5 . The artificial intelligence program 212 controls the first object 500 and the second object 510 included in the first captured image FA1 and the first object 500 and second object included in the second captured image FA2. The object 510 is analyzed using the learning data 241, and geographic feature information of a location where the first object 500 exists is identified according to the result of the analysis (S210).

The artificial intelligence program 212 analyzes the first moving direction of the first object 500 and the identified feature (or feature information) using the learning data 241 and determines the first destination of the first object 500. Predict (S130A).

The artificial intelligence program 212 analyzes the first moving method, the first moving speed, and the identified terrain feature using the learning data 214 and determines the first time for the first object 500 to reach the first destination. Predict (S135A).

According to embodiments, the first capture image FA1 and the capture image FA2 may be output from the same capture device or from different capture devices.

According to embodiments, since the database 240 stores the geographic feature information 243, the artificial intelligence program 212 provides first location information of the first capturing device 313 outputting the first captured image FA1. The second location information PI2 of the second capturing device 323 outputting PI1 and the second captured image FA2 is analyzed using the terrain feature information 243 and the first object 500 exists. A terrain feature (for example, a terrain feature corresponding to the terrain information TG1 and the feature information FI1) is identified (S210).

The moving speed when the first object 500 moves in a flat area and the moving speed when the first object 500 moves in a mountainous area with a steep slope in the same moving method are different from each other. Therefore, when the artificial intelligence program 212 predicts the moving speed of the first object 500, the artificial intelligence program 212 uses the identified terrain feature information to more accurately predict the moving speed of the first object 500. Predictable.

Since the steps S110 to S125 and S140 to S150 of FIG. 7 and the steps S110 to S125 and S140 to S150 of FIG. 4 are the same, description of the steps S110 to S125 and S140 to S150 is omitted. Also, steps S137 and S139 of FIG. 6 are applied between steps S135A and S140 of FIG. 7 .

According to embodiments, the first capture image FA1 and the capture image FA2 may be output from the same capture device or from different capture devices. The first captured image FA1 and the captured image FA2 may be images captured by the drone 700 .

The intrusion prediction method using an artificial intelligence algorithm (or artificial intelligence program 212) generates training data 241 by learning sample images (SPI) using the artificial intelligence algorithm (S110), and first captured images. (FA1, FB1, or FC1) is received, and the first object 500 included in the first captured image (FA1, FB1, or FC1) is identified using the learning data 241 (S115), and the second captured image Receiving the image (FA2, FB2, or FC2) and identifying the first object 500 included in the second captured image (FA2, FB2, or FC2) using the learning data 241 (S120), The movement of the first object 500 identified in each of the captured image FA1, FB1, or FC1 and the second captured image FA2, FB2, or FC2 is analyzed using the learning data 241 to determine the first object ( 500), the movement method, movement speed, and movement direction are acquired (S125), the movement direction is analyzed using the learning data 241 to predict the destination of the first object 500 (S130), and the movement The method and the moving speed are analyzed using the learning data 241 to predict the time at which the first object 500 will reach the destination (S135).

According to embodiments, the intrusion prediction method receives third captured images, identifies a first object 500 and a second object included in each of the third captured images using learning data 241, and 2 When the object is identified as the fence 400, it is determined whether the first object 500 is intruding through the fence 400 (S137), and the first object 500 is intruding into the fence 400. When it does, an alarm is generated and prediction information (IF) including the predicted destination and the predicted time is generated and transmitted to the terminal 600 (S140).

According to embodiments, the intrusion prediction method further learns sample sound signals (SDS) using an artificial intelligence algorithm to generate learning data 241 (S110), receives sound signals (SS), and generates sound signals It is determined whether (SS) are artificial sound signals using the learning data 241, and when the sound signals (SS) are determined to be the artificial sound signals, an intruder passes through the fence 400 through the first object 500. It is determined (YES in S137), an alert is generated, and prediction information (IF) including a predicted destination and a predicted time is generated and transmitted to the terminal 600 (S140).

According to the embodiments, the intrusion prediction method further learns sample electrical signals ELS using an artificial intelligence algorithm to generate learning data 241 (S110), receives electrical signals ES, and electrical signals It is determined whether (SS) are artificial electrical signals using the learning data 241 (S137), and when the electrical signals (ES) are determined to be the artificial electrical signals, the first object 500 is placed in the fence 400 It is determined as an intruder through (YES in S137), an alarm is generated, and prediction information (IF) including a predicted destination and a predicted time is generated and transmitted to the terminal 600 (S140).

According to the embodiments, the intrusion prediction method further learns sample vibration signals (VBS) using an artificial intelligence algorithm to generate learning data 241 (S110), receives vibration signals (VS), and generates vibration signals It is determined whether VS is artificial vibration signals using the learning data 241 (S137), and when the vibration signals VS are determined to be the artificial vibration signals, the first object 500 is moved to the fence 400. It is determined as an intruder through (YES in S137), an alarm is generated, and prediction information (IF) including a predicted destination and a predicted time is generated and transmitted to the terminal 600 (S140).

According to an embodiment, the artificial intelligence program 212 may perform the second object 510 included in the first captured image FA1, FB1, or FC1 and the second object 510 included in the second captured image FA2, FB2, or FC2. The object 510 is analyzed using the learning data 241 to identify the feature at the location where the first object 500 exists (S210), and the moving direction of the obtained first object 500 and the identified feature The water is analyzed using the learning data 241 to predict the destination of the first object 500 (S130A).

The artificial intelligence program 212 analyzes the acquired movement method, the acquired movement speed, and the identified terrain feature using the learning data 241 to predict the time for the first object 500 to reach the destination (S135A). ).

According to an embodiment, the artificial intelligence program 212 analyzes location information where the same photographing device outputting the first captured image (FA1, FB1, or FC1) and the second captured image (FA2, FB2, or FC2) is installed. The first object 500 is identified, and the destination of the first object 500 is predicted by analyzing the obtained movement direction and the identified feature using the learning data 241 (S130A). .

The artificial intelligence program 212 analyzes the acquired movement method, the acquired movement speed, and the identified terrain feature using the learning data 241 to predict the time for the first object 500 to reach the destination (S135A). ).

According to an embodiment, the artificial intelligence program 212 may generate the first location information PI1 of the first capturing device 313 outputting the first captured image FA1, FB1, or FC1 and the second captured image FA2 or FB2. , or FC2) is analyzed to analyze the second location information (PI2) of the second photographing device 323 to identify the feature where the first object 500 exists, and the obtained movement direction and the identified feature are analyzed. The destination of the first object 500 is predicted by analyzing the learning data 241 (S130A).

The artificial intelligence program 212 analyzes the acquired movement method, the acquired movement speed, and the identified terrain feature using the learning data 241 to predict the time for the first object 500 to reach the destination (S135A). ).

The artificial intelligence program 212 receives the third captured image, identifies the first object 500 included in the third captured image using the learning data 241 (S145), and performs the second captured image (FA2, A new movement method of the first object 500, a new movement speed, and A new movement direction is obtained (S145), the new movement direction is analyzed using the learning data 241 to predict a new destination of the first object 500 (S145), and the new movement method and new movement speed are determined from the learning data By analyzing using step 241, the time at which the first object 500 will arrive at a new destination is predicted (S145).

The artificial intelligence program 212 changes the previous destination to a new destination, changes the previously predicted first time to the newly predicted time, and generates prediction information including the new destination and the new time, so that the terminal 600 It is transmitted to (S150).

Although the present invention has been described with reference to the embodiments shown in the drawings, this is only exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the attached claims.

100: intrusion alert system
200: control device using artificial intelligence
311, 321, 331, 341, 351, 361, 371, 381, 391: sensor
313, 323, 333, 343, 353, 363, 373, 383, 393: photographing device
400: fence
410: facility
500: object, person

Claims (6)

filming device;
a sensor that detects vibration; and
Including a control device using artificial intelligence,
The control device using the artificial intelligence,
database; and
contains a processor;
the processor,
Learning sample images using an artificial intelligence algorithm to generate image signal learning data and storing the image signal learning data in the database;
Learning sample vibration signals using the artificial intelligence algorithm to generate vibration signal learning data and storing the vibration signal learning data in the database;
Receiving a first captured image transmitted from the photographing device and identifying an object included in the first captured image using the image signal learning data;
Receiving a second captured image transmitted from the photographing device and identifying the object included in the second captured image using the image signal learning data;
Receiving vibration signals output from the sensor, determining whether the vibration signals are artificial vibration signals using the vibration signal learning data;
When the vibration signals are determined to be the artificial vibration signals, the object is determined as an intruder through the fence;
When the object is determined to be the intruder, the movement of the object identified in each of the first and second captured images is analyzed using the image signal learning data to determine the moving method, moving speed, and obtain a direction of movement, including an azimuth and an angular range;
Analyzing the moving direction using the video signal learning data to predict whether the object will go to a destination through one of a plurality of moving paths;
predicting a time when the object will reach the destination by analyzing the moving method and the moving speed using the video signal learning data;
An intrusion detection system for generating prediction information including the destination and a time to reach the destination. The method of claim 1, wherein the control device using artificial intelligence,
and a display device for receiving and displaying the first captured image and the second captured image under the control of the processor. According to claim 1,
The intrusion boundary system of claim 1, wherein the photographing device is a CCTV camera, a network camera, a pan-tilt-zoom (PTZ) camera, a thermal imaging camera, or a thermal imaging monitoring device. a first photographing device;
a second photographing device;
a sensor that detects vibration; and
Including a control device using artificial intelligence,
The control device using the artificial intelligence,
database; and
contains a processor;
the processor,
Learning sample images using an artificial intelligence algorithm to generate image signal learning data and storing the image signal learning data in the database;
Learning sample vibration signals using the artificial intelligence algorithm to generate vibration signal learning data and storing the vibration signal learning data in the database;
Receiving a first captured image transmitted from the first capturing device and identifying an object included in the first captured image using the image signal learning data;
Receiving a second captured image transmitted from the second capturing device and identifying the object included in the second captured image using the image signal learning data;
Receiving vibration signals output from the sensor, determining whether the vibration signals are artificial vibration signals using the vibration signal learning data;
When the vibration signals are determined to be the artificial vibration signals, the object is determined as an intruder through the fence;
When the object is determined to be the intruder, the movement of the object identified in each of the first and second captured images is analyzed using the image signal learning data to determine the moving method, moving speed, and obtain a direction of movement, including an azimuth and an angular range;
Analyzing the moving direction using the video signal learning data to predict whether the object will go to a destination through one of a plurality of moving paths;
predicting a time when the object will reach the destination by analyzing the moving method and the moving speed using the video signal learning data;
An intrusion detection system for generating prediction information including the destination and a time to reach the destination. database;
sensor interface;
shooting device interface; and
contains a processor;
the processor,
Learning sample images using an artificial intelligence algorithm to generate image signal learning data and storing the image signal learning data in the database;
Learning sample vibration signals using the artificial intelligence algorithm to generate vibration signal learning data and storing the vibration signal learning data in the database;
Receiving a first captured image transmitted from a photographing device and identifying an object included in the first captured image using the image signal learning data;
Receiving a second captured image transmitted from the photographing device and identifying the object included in the second captured image using the image signal learning data;
Receiving vibration signals output from a sensor that detects vibration through the sensor interface, and determining whether the vibration signals are artificial vibration signals using the vibration signal learning data;
When the vibration signals are determined to be the artificial vibration signals, the object is determined as an intruder through the fence;
When the object is determined to be the intruder, the movement of the object identified in each of the first and second captured images is analyzed using the image signal learning data to determine the moving method, moving speed, and obtain a direction of movement, including an azimuth and an angular range;
Analyzing the moving direction using the video signal learning data to predict whether the object will go to a destination through one of a plurality of moving paths;
predicting a time when the object will reach the destination by analyzing the moving method and the moving speed using the video signal learning data;
generating prediction information including the destination and a time to reach the destination;
receiving and interpreting the vibration signals from the sensor interface, and generating a control signal for controlling the photographing device according to an analysis result;
wherein the photographing device interface receives the control signal from the processor and transmits it to the photographing device. database;
sensor interface;
shooting device interface; and
contains a processor;
the processor,
Learning sample images using an artificial intelligence algorithm to generate image signal learning data and storing the image signal learning data in the database;
Learning sample vibration signals using the artificial intelligence algorithm to generate vibration signal learning data and storing the vibration signal learning data in the database;
Receiving a first captured image transmitted from a first capturing device and identifying an object included in the first captured image using the image signal learning data;
Receiving a second captured image transmitted from a second capturing device and identifying the object included in the second captured image using the image signal learning data;
Receiving vibration signals output from a sensor that detects vibration through the sensor interface, and determining whether the vibration signals are artificial vibration signals using the vibration signal learning data;
When the vibration signals are determined to be the artificial vibration signals, the object is determined as an intruder through the fence;
When the object is determined to be the intruder, the movement of the object identified in each of the first and second captured images is analyzed using the image signal learning data to determine the moving method, moving speed, and obtain a direction of movement, including an azimuth and an angular range;
Analyzing the moving direction using the video signal learning data to predict whether the object will go to a destination through one of a plurality of moving paths;
predicting a time when the object will reach the destination by analyzing the moving method and the moving speed using the video signal learning data;
generating prediction information including the destination and a time to reach the destination;
receiving and interpreting the vibration signals from the sensor interface, and generating a control signal for controlling the first photographing device and the first photographing device according to an analysis result;
wherein the photographing device interface receives the control signal from the processor and transmits the control signal to the first photographing device and the second photographing device.

Images