Security monitoring system, security monitoring method, and security monitoring program

Abstract

To realize security monitoring in a manner personalized to the unique environment and circumstances of each monitored area. [Solution] A security monitoring system (100) comprising an imaging unit (1), an image analysis unit (31), a monitoring necessity information storage unit (32), a monitoring information notification unit (33), and a monitoring necessity information update unit (35), wherein the image analysis unit (31) analyzes the monitoring image using artificial intelligence functions to detect objects requiring confirmation, and further determines whether the objects requiring confirmation are identical to the objects requiring monitoring included in the monitoring necessity information, wherein the monitoring information notification unit (33) outputs warning information for the objects requiring confirmation if it is determined that the objects requiring confirmation are not identical, and wherein the monitoring necessity information update unit (35), when a user (5) inputs a judgment result regarding whether or not to continue monitoring the objects requiring confirmation output as warning information, dynamically updates the contents of the monitoring necessity information to match the judgment result.

Description

[Technical Field]

[0001] The present invention relates to a security monitoring system, a security monitoring method, and a security monitoring program. More specifically, the present invention relates to a security monitoring technology that utilizes artificial intelligence (AI), and relates to a security monitoring system, a security monitoring method, and a security monitoring program that can perform security monitoring in a manner optimized for each monitored area when security monitoring tailored to the unique environment and circumstances of each monitored area, such as the area surrounding a specific private residence or a specific cultural heritage building. [Background technology]

[0002] For example, in the conventional security monitoring system disclosed in Patent Document 1, images captured by a camera are analyzed to detect objects that meet certain criteria, such as abnormal behavior or suspicious individuals. When an object determined to be a suspicious individual or object is discovered, a notification is sent to the "user" or "system administrator" of the security monitoring system. In this specification, the term "user" refers to a person who enjoys the benefits of the security monitoring system or their representative. In the case of a security monitoring system for a private residence, the "user" would be the resident or owner of the residence, or a person entrusted by such a person with the actual monitoring work. In this specification, the term "system administrator" refers to the information processing engineer responsible for maintenance and management tasks to keep the system operating normally or in a more desirable manner.

[0003] However, the security monitoring system disclosed in Patent Document 1 is basically a system that performs routine monitoring and warnings based on a predetermined algorithm set as a default. Therefore, it is difficult for this security monitoring system to flexibly respond to the unique environments and circumstances that differ for each "individual monitored area," each of which is subject to different unique conditions. Specifically, it is difficult to always ensure that a specific target that should be detected only under the unique environment and circumstances of each monitored area (for example, a case in which an ordinary vehicle that is not generally considered to be highly suspicious but is unrelated to the area repeatedly appears in the area) is not omitted from the detection targets.

[0004] To address this issue, Patent Document 1 discloses that information about detected targets can be recorded in a storage device and used for reanalysis or reference as needed. However, in order to continue to accurately determine the level of suspiciousness in accordance with the unique environments and circumstances of each monitored area, a "system administrator" responsible for maintaining and managing the system must frequently and irregularly perform manual operations such as retraining the artificial intelligence (AI) or rewriting the program to correct erroneous detections each time a false positive occurs. This excessive workload on the "system administrator" has been a practical obstacle to achieving security monitoring tailored to the unique environments and circumstances of each monitored area in the field of security monitoring using artificial intelligence (AI). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-294921 Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention aims to provide a security monitoring system that uses the image analysis functions of artificial intelligence (AI) to detect targets for monitoring, and in cases where security monitoring tailored to the unique environment and circumstances of each individual monitored area, such as a specific private residence or the area surrounding a specific cultural heritage building, achieves detailed security monitoring in an optimally personalized manner tailored to the environment and circumstances without placing an excessive burden on the "system administrator." [Means for solving the problem]

[0007] The inventors of the present application have conceived and completed the present invention, which is based on the idea that in a security monitoring system that utilizes the advanced image analysis functions of artificial intelligence (AI), the results of the user's judgment on the selection of monitoring targets can be continuously incorporated into the system in a cyclical process unique to the present invention, thereby enabling the system to constantly provide a "monitoring function personalized for each user" without relying on an excessive burden on a "system administrator." Specifically, the present invention provides the following security monitoring systems, etc.

[0008] (1) A security monitoring system for monitoring a specific area to be monitored, comprising: a photographing unit for capturing monitoring images of the area to be monitored; an image analysis unit for performing image analysis of the monitoring images using an image analysis function of artificial intelligence; a monitoring necessity information storage unit for storing monitoring necessity information including a list of monitoring necessity targets specific to the area to be monitored; a monitoring information notification unit for notifying a user of the security monitoring system of the monitoring information; and a monitoring necessity information update unit, wherein the image analysis unit analyzes the monitoring images and detects, as a monitoring necessity target, a monitoring necessity target that is determined to have a certain level of suspiciousness or higher; and a verification necessity target detection unit for detecting, as a monitoring necessity target, a monitoring necessity target that is determined to have a certain level of suspiciousness or higher, and a comparison between the verification necessity target and the monitoring necessity target included in the monitoring necessity information. a monitoring necessity determination means for determining identity, wherein the monitoring information notification unit has a monitoring information selection and output means for outputting warning information for the target to be confirmed if it is determined that the target to be confirmed is the same as the target to be monitored, and for outputting caution information for the target to be confirmed if it is determined that the target to be confirmed is not the same as the target to be monitored, and the monitoring necessity information update unit has a monitoring necessity information update means for dynamically updating the content of the monitoring necessity information to match the judgment result when a judgment result is input by a user of the security monitoring system regarding whether or not to continue monitoring the target to be confirmed that has been output as the caution information.

[0009] According to the security monitoring system of (1), in the field of security monitoring utilizing artificial intelligence (AI), when security monitoring tailored to the specific environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, detailed security monitoring can be realized in an optimally personalized manner tailored to the environment and circumstances without relying on an excessive burden on the "system administrator."

[0010] (2) The security monitoring system described in (1), wherein the monitoring necessity information also includes a list of objects that do not need to be monitored that are specific to the monitored area, and the monitoring information notification unit does not output information about the object that needs to be confirmed if it determines that the object that needs to be confirmed is the same as the object that does not need to be monitored.

[0011] According to the security monitoring system of (2), even if the artificial intelligence (AI) detects a person or object that is initially deemed suspicious, if the "user" determines that the detected object "does not require monitoring" based on his / her own criteria, the "user" can quickly remove the detected object from the list of objects requiring monitoring through an easy-to-perform operation without relying on an excessive burden on the "system administrator." This allows for more efficient and thorough optimization of the system through dynamic updates of the monitoring necessity information.

[0012] (3) A security monitoring system as described in (1) or (2), further comprising a recording unit that records the surveillance images as stored images, and an image search unit that searches the stored images to extract specific surveillance targets, wherein the image analysis unit has a suspiciousness judgment standard change means that automatically changes the criteria for judging the suspiciousness of the surveillance target depending on the results of analyzing the stored images of the surveillance target extracted by the image search unit.

[0013] According to the security monitoring system of (3), in the security monitoring system of (1) or (2), specific images that can contribute to real-time judgment of suspiciousness can be extracted from past monitoring images (accumulated images) stored in the recording unit, and the analysis results can also be automatically contributed to the judgment of suspiciousness in real-time detection of a target of investigation. This makes it possible to make more advanced and complex judgments that incorporate not only the environment and situation in the monitored area, but also the history of past events and changes in events over time up to the present, and realizes security monitoring in a more detailed and personalized manner that is tailored to the actual conditions of the monitored area.

[0014] (4) A security surveillance system that monitors a surveillance area within a specific range, comprising: a photographing unit that acquires surveillance images of the surveillance area; a recording unit that records the surveillance images as accumulated images; an image search unit that searches the accumulated images and extracts specific surveillance targets; an image analysis unit that performs image analysis of the surveillance images using artificial intelligence; and a surveillance information notification unit that notifies a user of the security surveillance system of surveillance information, wherein the image analysis unit has a verification target detection means that analyzes the surveillance images and detects surveillance targets that are determined to have a certain level of suspiciousness or higher as surveillance targets, and a suspiciousness determination standard change means that automatically changes the criteria for determining the suspiciousness of the surveillance targets depending on the results of analyzing the accumulated images of the surveillance targets extracted by the image search unit.

[0015] According to the security monitoring system of (4), when security monitoring tailored to the unique environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, specific images that can contribute to a real-time judgment of suspiciousness can be extracted from past monitoring images stored in the recording unit, and the analysis results can be automatically used to judge the suspiciousness of targets in real-time. This makes it possible to make sophisticated and complex judgments that incorporate the history of past events and the chronological changes in events up to the present, and to achieve detailed security monitoring in a manner that is optimally personalized to the environment and circumstances, without relying on an excessive burden on the "system administrator."

[0016] (5) A security monitoring method for monitoring a monitoring target area within a specific range, comprising: a photographing step in which a photographing unit photographs the monitoring target area and obtains a monitoring image; a confirmation target detection step in which an image analysis unit performs image analysis of the monitoring image using artificial intelligence to detect, as a monitoring target, a monitoring target that is determined to have a certain level of suspiciousness or higher; a monitoring necessity determination step in which the image analysis unit determines whether the confirmation target is identical to a monitoring target stored in the security monitoring system; a monitoring information notification step in which a monitoring information update unit outputs warning information for the confirmation target if it is determined that the confirmation target is identical to the monitoring target, and outputs caution information for the confirmation target if it is determined that the confirmation target is not identical to the monitoring target; and a monitoring necessity information update step in which, when a user of the security monitoring system inputs a judgment result regarding whether or not to continue monitoring the confirmation target output as the caution information, the monitoring necessity information update unit dynamically updates the content of the monitoring necessity information to match the judgment result.

[0017] According to the security monitoring method (5), in the field of security monitoring using artificial intelligence (AI), when security monitoring tailored to the specific environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, detailed security monitoring can be realized in an optimally personalized manner tailored to the environment and circumstances without relying on an excessive burden on the "system administrator."

[0018] (6) A security monitoring method as described in (5), in which the monitoring necessity information also includes a list of objects that do not need to be monitored that are specific to the monitored area, and the monitoring information output step unit does not output information about the object that needs to be confirmed if it is determined that the object that needs to be confirmed is the same as the object that does not need to be monitored.

[0019] According to the security monitoring method of (6), even if an object is detected by artificial intelligence (AI) as being initially suspicious, if the "user" determines that the object "does not require monitoring" based on his / her own criteria, the "user" can quickly remove the object from the list of objects requiring monitoring by performing an easy operation without placing an excessive burden on the "system administrator." This allows for more efficient and thorough optimization of the system by dynamically updating the content of the monitoring necessity information.

[0020] (7) A security monitoring method as described in (5) or (6), further comprising a recording step in which a recording unit records the surveillance images as accumulated images, and an accumulated image search step in which an image search unit searches the recorded surveillance images to extract accumulated images of specific surveillance targets, wherein in the target detection step, the criteria for determining the suspiciousness of the surveillance target are automatically updated according to the results of analyzing the accumulated images extracted in the accumulated image search step, and the target to be confirmed is detected based on the updated criteria.

[0021] According to the security monitoring method of (7), in the security monitoring method of (5) or (6), specific images that can contribute to real-time judgment of suspiciousness can be extracted from past surveillance images stored in the recording unit, and the analysis results can also be automatically contributed to the judgment of suspiciousness in real-time detection of a target of investigation. This makes it possible to make more advanced and complex judgments that incorporate not only the environment and situation in the monitored area, but also the history of past events and changes in events over time up to the present, and realizes security monitoring in a more detailed and personalized manner that is tailored to the actual conditions of the monitored area.

[0022] (8) A security surveillance method for monitoring a surveillance area within a specific range, comprising: a photographing step in which a photographing unit photographs the surveillance area to obtain surveillance images; a recording step in which a recording unit records the surveillance images as accumulated images; an accumulated image search step st3 in which an image search unit searches the recorded surveillance images and extracts the accumulated images of specific surveillance targets; and a target detection step in which an image analysis unit performs image analysis of the surveillance images using artificial intelligence to detect surveillance targets that are determined to have a certain level of suspiciousness or higher as targets requiring confirmation, wherein in the target detection step, the criteria for determining the suspiciousness of the surveillance target are automatically updated according to the results of analyzing the accumulated images extracted in the accumulated image search step st3, and the target requiring confirmation is detected based on the updated criteria.

[0023] According to the security monitoring method of (8), when security monitoring tailored to the unique environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, specific images that can contribute to a real-time judgment of the degree of suspiciousness can be extracted from past monitoring images stored in the recording unit, and the analysis results can be automatically used to judge the degree of suspiciousness in the real-time detection of a target requiring investigation. This makes it possible to make sophisticated and complex judgments that incorporate the history of past events and the chronological changes in events up to the present, and to realize detailed security monitoring in a manner that is optimally personalized to the environment and circumstances, without relying on an excessive burden on the "system administrator."

[0024] (9) A security monitoring program that monitors a monitoring target area within a specific range, the security monitoring program causing a computer to execute a security monitoring method that includes: a verification target detection step that uses artificial intelligence to perform image analysis of monitoring images to detect, as a target requiring confirmation, a target for monitoring that is determined to have a certain level of suspiciousness or higher; a monitoring necessity determination step that determines whether the target for monitoring is identical to a target for monitoring that is stored in the security monitoring system; a monitoring information notification step that outputs warning information for the target for monitoring if it is determined that the target for monitoring is identical to the target for monitoring, and outputs caution information for the target for monitoring if it is determined that the target for monitoring is not identical to the target for monitoring; and a monitoring necessity information update step that, when a user of the security monitoring system inputs a judgment result regarding whether or not to continue monitoring the target for monitoring that has been output as caution information, dynamically updates the content of the monitoring necessity information to match the judgment result.

[0025] According to the security monitoring program of (9), when security monitoring tailored to the specific environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, detailed security monitoring can be performed in an optimally personalized manner tailored to the environment and circumstances, with easy operation by the "user" himself, without relying on an excessive burden on the "system administrator."

[0026] (10) A security monitoring program for monitoring a monitored area within a specific range, which includes an accumulated image search step for searching through recorded surveillance images to extract accumulated images of specific monitored targets, and an object to be confirmed detection step for performing image analysis of the surveillance images using artificial intelligence to detect monitored targets that are determined to have a certain level of suspiciousness or higher as objects to be confirmed, and which causes a computer to execute a security monitoring method in which, in the object to be confirmed detection step, the criteria for determining the level of suspiciousness are automatically changed depending on the results of analyzing the accumulated images.

[0027] According to the security monitoring program of (10), when security monitoring tailored to the unique environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, specific images that can contribute to a real-time judgment of the degree of suspiciousness can be extracted from past monitoring images stored in the recording unit, and the analysis results can be automatically used to judge the degree of suspiciousness in the real-time detection of a target requiring investigation. This makes it possible to make sophisticated and complex judgments that incorporate the history of past events and changes in events over time up to the present, and to realize detailed security monitoring in a manner that is optimally personalized to the environment and circumstances, without relying on an excessive burden on the "system administrator." [Effects of the Invention]

[0028] According to the present invention, when security monitoring tailored to the unique environment and circumstances of each individual monitored area is required, such as the area surrounding a specific private residence or a specific cultural heritage building, detailed security monitoring can be achieved in a manner that is optimally personalized to suit the environment and circumstances, without relying on an excessive burden on the ``system administrator.'' [Brief explanation of the drawings]

[0029] [Figure 1] 1 is a block diagram showing a configuration of a security monitoring system according to the present invention; [Figure 2] 3 is a flowchart showing the flow of operations in the security monitoring method of the present invention. [Figure 3] 1 is a conceptual diagram showing the basic design concept, configuration, and operation of an image analysis unit provided in the security monitoring system of the present invention. [Figure 4] 1 is a diagram schematically illustrating an example of a specific overall configuration of a security monitoring system according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0030] <Security monitoring system> The best mode for carrying out the present invention will be described below with reference to the accompanying drawings. The security monitoring system of the present invention (hereinafter simply referred to as the "security monitoring system") is a monitoring system for maintaining security within a monitored area within a specific range.

[0031] The "security monitoring system" first notifies the "user" of the system, who is the beneficiary of the monitoring function, of "targets requiring surveillance" automatically detected by artificial intelligence (AI) image analysis as "warning information" or "alert information" that prompts the user to monitor the target. The "security monitoring system" then allows the "user" to determine for himself or herself whether the notified "targets requiring surveillance (warning information)" require continued surveillance. If necessary, the "user" can easily register the "targets requiring surveillance" on a "targets requiring surveillance" list or remove them from future surveillance ("register to no-survey list") using a smartphone or other device. This allows the "user" of the "monitoring system" to enjoy continuous surveillance benefits that are more precisely tailored to the "user's" security needs than conventional security monitoring systems, without relying on the support of a "system administrator." The specific configuration of the "security monitoring system" that is capable of implementing this unique operation is described in detail below.

[0032] [Overall configuration] The basic configuration of a security monitoring system 100, which is an example of a preferred embodiment of the "security monitoring system," is as shown in the block diagram of Figure 1. The security monitoring system 100 is configured to include a photographing unit 1 and a processing unit 3. The processing unit 3 is configured to include at least an image analysis unit 31, a monitoring necessity information storage unit 32, a monitoring information notification unit 33, and a monitoring necessity information update unit 35. In addition to the above-mentioned partial configurations, the security monitoring system 100 preferably further includes a recording unit 2, and the processing unit 3 further includes an image search unit 34.

[0033] As for the user terminal 4 used by user 5 (see FIG. 4), who is the "user" of security monitoring system 100, the present invention can be implemented by, for example, using any general-purpose information processing terminal (such as a smartphone owned before starting to use the security monitoring system) that user 5 has owned before acquiring the system, which is the invention, as needed, without any particular restrictions. Therefore, the following description will be given assuming that this user terminal 4 is not included in the minimum essential components of the "security monitoring system" of the present invention.

[0034] As an example of the specific arrangement of the components of the security monitoring system 100, as shown in FIG. 4 , a photographing unit 1 (e.g., a digital IP camera for video recording) installed in a monitored area (e.g., an area where a monitored object 6 (e.g., the home of a user 5) is located) is connected to a processing unit 3 located on the cloud, and a user terminal 4 portable by the user 5 via a network for communication. In this case, the communication means constituting the network for communicatively connecting the above components is not particularly limited and can be any of various wired or wireless telecommunications lines. This communication means can be a wired connection using a dedicated communication cable, or it can be a communication means via the Internet or any of various telephone lines. Furthermore, these communication means are preferably configured to be capable of communication using a communication protocol compliant with the ONVIF (Open Network Video Interface Forum) standard, but are not limited thereto and may be configured to be capable of communication using other communication protocols, such as a proprietary communication protocol.

[0035] [Photography Department] The photographing unit 1 is a device that photographs the area to be monitored and acquires a surveillance image. The camera that constitutes the photographing unit 1 may be any camera that can output the photographed surveillance image as digital image data onto the network that constitutes the security monitoring system 100 so that the captured surveillance image can be processed by the arithmetic processing unit 3. Specifically, a digital IP camera for video recording that has various existing video recording functions can be preferably used as the camera that constitutes the photographing unit 1.

[0036] Furthermore, it is more preferable to use a camera with a wide-angle lens and zoom function as the camera that constitutes the imaging unit 1 so that a wide monitoring area can be covered with one camera. Alternatively, it is also preferable to use a PTZ camera that can be remotely controlled to pan, tilt, and zoom. Such a camera allows a wider area to be monitored with one camera. In other words, by using such cameras, the entire monitoring area can be covered with fewer cameras.

[0037] Furthermore, particularly when the area to be monitored is outdoors, or even indoors where it is expected to be dark, it is preferable to use a camera equipped with a high-sensitivity sensor capable of capturing images in low-light environments as the camera that constitutes the imaging unit 1. There are outdoor-compatible security cameras on the market that are equipped with high-sensitivity sensors that can capture subjects in color even with a weak light source with a minimum illumination of about 0.009 lux, and by using a camera equipped with such a high-sensitivity sensor as the camera that constitutes the imaging unit 1, the security monitoring system 100 can be made to be sufficiently effective even when monitoring at night or in dark places.

[0038] Furthermore, among the cameras constituting the image capture unit 1, it is particularly effective to use cameras that monitor entrances and exits of the monitored area (for example, the front door of a private residence) equipped with a facial recognition function that can recognize the faces of people entering and exiting the entrance. The facial recognition function may be built into the image capture unit 1 or into the processing unit 3. By configuring the security monitoring system 100 in this way, it is possible for the security monitoring system 100 to distinguish between people who normally reside within the monitored area and other people, and to take special measures against people on a blacklist or whitelist. This makes it possible to achieve more adaptable and detailed monitoring effects that are tailored to the unique circumstances of each monitored area.

[0039] The camera constituting the image capturing unit 1 may be a general monocular camera capable of continuously capturing a three-dimensional space as a two-dimensional image. Alternatively, a 3D camera capable of directly acquiring three-dimensional position information within the space that is the imageable area may be used as the camera constituting the image capturing unit 1. When the monocular camera is used as the camera constituting the image capturing unit 1, it is more preferable that the security monitoring system 100 further includes a coordinate setting unit (not shown in FIG. 1 ) that sets identifiable coordinates by relating a position in the “monitoring image” to an actual position within the three-dimensional space that is the imageable area. The coordinate setting unit may be built into the image capturing unit 1 or may be incorporated into the arithmetic processing unit 3. By providing a coordinate setting unit capable of generating three-dimensional coordinates from two-dimensional image information, it is possible to detect the movement of a person or an object with high accuracy even from images containing only two-dimensional information that can be acquired by an inexpensive monocular camera, without introducing an expensive 3D camera.

[0040] [Recording section] The recording unit 2 is a device that records past images of the monitored area and stores them as accumulated images. The recording unit 2 stores the monitoring images (previously recorded images) captured by the imaging unit 1 as digital image data so that they can be processed by the processing unit 3 after a certain period of time has passed, and has the function of extracting the necessary portions as digital data as needed. Specifically, a digital recorder having various existing video recording functions and a data storage device can be used as the recorder that constitutes the recording unit 2. The recording unit 2 may also be a recording device that integrates a digital recording device and a data storage device, or it may be a recording system in which devices having the above functions are distributed over a network.

[0041] [Calculation processing unit] The arithmetic processing unit 3 analyzes surveillance images using image analysis technology possessed by artificial intelligence (AI), detects highly suspicious objects, and notifies security-related information to the user terminal 4 carried by the user 5. In order to perform such functions, the arithmetic processing unit 3 essentially comprises an image analysis unit 31, a surveillance necessity information storage unit 32, a surveillance information notification unit 33, and a surveillance necessity information update unit 35.

[0042] Furthermore, when the security monitoring system 100 is configured to include a recording unit 2, it is preferable that the calculation processing unit 3 also includes an image search unit 34, thereby enabling the security monitoring system 100 to automatically contribute the results of analysis of past surveillance images (accumulated images) to the real-time artificial intelligence (AI) judgment of the degree of suspiciousness.

[0043] The arithmetic processing unit 3 is a known information processing device of any of various types equipped with hardware such as a CPU, memory, and communication unit, and can be configured as a computer system that appropriately combines various information processing equipment and electronic devices that can perform the functions required of each of the above units (image analysis unit 31, monitoring necessity information storage unit 32, monitoring information notification unit 33, image search unit 34, and monitoring necessity information update unit 35).The arithmetic processing unit 3 only needs to be connected to the photographing unit 1, recording unit 2, and user terminal 4 via the various communication networks described above so as to be able to send and receive data to and from each other.

[0044] Furthermore, in the security monitoring system 100, the computer system that constitutes the arithmetic processing unit 3 as described above can be placed on the cloud as a server, and this can be configured to be shared by multiple clients (such as the individual cameras that constitute the photographing unit 1 and multiple smartphones used as user terminals 4). Furthermore, the arithmetic processing unit 3 can automatically execute each operation of the security monitoring system 100 according to instructions from a computer program (security monitoring program) according to the present invention.

[0045] [Image analysis section] The image analysis unit 31 has a "detection means for target to be confirmed" that detects targets of surveillance (people, objects, vehicles, etc.) that the artificial intelligence (AI) has determined to be suspicious to a certain level or above by analyzing surveillance images using the object recognition function and behavior analysis function of the artificial intelligence (AI), as "targets to be confirmed", and a "monitoring necessity determination means" that determines whether the detected "targets to be confirmed" are identical to the "targets to be monitored" included in the monitoring necessity information stored in the monitoring necessity information storage unit 32.

[0046] Furthermore, as mentioned above, the security monitoring system 100 can be configured to include a recording unit 2 and an image search unit 34. In this case, the image analysis unit 31 can be configured to have a "suspiciousness judgment criteria changing means" that automatically changes the criteria for judging the suspiciousness of a monitored object based on the results of analyzing the accumulated images of the monitored object extracted by the image search unit 34. This enables artificial intelligence (AI) to make advanced and complex judgments that incorporate past events and the history of changes in events over time up to the present, thereby enabling detailed security monitoring to be realized in a manner that is optimally personalized to suit the environment and situation.

[0047] (Means for changing criteria for determining suspiciousness) The suspiciousness judgment criterion changing means can be configured by various kinds of arithmetic processing means having a function of performing arithmetic processing to automatically change the judgment criteria (e.g., the suspiciousness score threshold) when artificial intelligence (AI) judges the suspiciousness of a monitoring target, in accordance with the results of analyzing the accumulated images of the monitoring target extracted by the image search unit 34. Specific details of this judgment criterion changing process will be described later in the description of an embodiment of the "security monitoring method" of the present invention.

[0048] (Method for detecting objects requiring confirmation) The identification target detection means can be configured using various known image analysis means capable of extracting the shape, size, and category of a "monitoring target" and its specific movements and actions through image analysis. Specific examples of such image analysis means include various machine learning image analysis devices with neural networks (so-called "image recognition devices using deep learning technology"). The identification target detection means constituting the image analysis unit 31 can be configured by installing a machine learning model trained to detect the target to be monitored (people, objects, vehicles, etc.) in the image analysis device. Specific examples of image recognition technology using deep learning are also disclosed below. "Deep Learning and Image Recognition, Operations Research" (http: / / www.orsj.o.jp / archive2 / or60-4 / or60_4_198.pdf)

[0049] Furthermore, one example of a technical means for mechanically recognizing the details of the actions and movements of a person in a surveillance image is a well-known image analysis technology called "OpenPose." As disclosed by the present inventor in Japanese Patent No. 6534499, "OpenPose" can be used to extract skeletal information of a person in the image and analyze the position and velocity of each feature point that constitutes the skeleton, thereby recognizing the person's actions in the image. Furthermore, image analysis technology such as "OpenPose" can extract the hand movements of the person being monitored and the movements of objects in the surveillance image as independent pieces of information from the "surveillance image." This makes it possible to comprehensively assess the suspiciousness of a person's actions toward an "object," such as when a person grabs an object and then carries it away.

[0050] (Method for determining whether monitoring is necessary) The monitoring necessity determination means can be configured using various known image analysis means (such as the above-mentioned "image recognition device using deep learning technology") that can compare an image of a "subject to be confirmed" with an image of a "subject to be monitored," or an image of a "subject to be confirmed" with an image of a "subject not to be monitored," and automatically determine the identity (or non-identity) of the two.

[0051] [Monitoring necessity information storage section] The monitoring necessity information storage unit 32 has a function of storing "monitoring necessity information" that serves as a criterion for determining whether monitoring is necessary in a monitored area within a specific range. The monitoring necessity information includes at least a list of "monitoring-required objects," which are objects that require monitoring in the monitored area where monitoring is actually performed. In addition, it is preferable that the monitoring necessity information also includes a list of "monitoring-unnecessary objects," which are objects that do not require monitoring in the monitored area. The monitoring necessity information storage unit 32 can be configured using various information storage means (storage devices) that can register information that has been organized into a database of each of these pieces of information.

[0052] [Monitoring information notification department] The monitoring information notification unit 33 has a function of selectively notifying the user 5 of multiple types of monitoring information with different required alert levels, i.e., "alert information" and "warning information." If the "target requiring confirmation" detected by the image analysis unit 31 is determined to be the same as any of the "targets requiring monitoring" stored in the monitoring necessity information storage unit 32, the monitoring information notification unit 33 selectively outputs "alert information" for the "target requiring confirmation." On the other hand, if the "target requiring confirmation" is determined to be different from the "target requiring monitoring," the monitoring information notification unit 33 selectively outputs "warning information" for the "target requiring confirmation." The monitoring information notification unit 33 outputs the monitoring information ("alert information" or "warning information") selectively output in this manner to the user terminal 4 carried by the user 5.

[0053] In this way, the security monitoring system 100 clearly distinguishes between "alert information" and "warning information," and selectively outputs information requiring different alert levels depending on the monitored object and notifies the user 5. In particular, by selecting "warning information," notifying the user 5 and asking for a decision on whether to continue monitoring, and then cyclically executing a process in which the result of that decision is reliably reflected in the result of the artificial intelligence (AI)'s suspiciousness determination from the next time onwards through an easy-to-implement operation by the user 5, the system progresses the process while always optimally combining the advanced arithmetic processing capabilities of the artificial intelligence (AI) with the user's judgment ability in line with the actual situation of the monitored area, which is a technical feature that differs from conventional security monitoring systems equipped with artificial intelligence (AI).

[0054] Furthermore, if the monitoring necessity information stored in the monitoring necessity information memory unit 32 also includes a list of objects that do not need to be monitored that are specific to the monitored area, the monitoring information notification unit 33 can prevent the output of either warning information or caution information for the object that needs to be checked if it is determined that the object that needs to be checked detected by the image analysis unit 31 is the same as an object that does not need to be monitored.

[0055] [Image Search Section] The image search unit 34 has a function of extracting a specific surveillance target by searching the recorded surveillance images (accumulated images) that have been recorded and stored by the recording unit 2. The image analysis unit 31 automatically changes the criteria for determining the suspiciousness of the surveillance target by the suspiciousness determination criteria changing means described above, depending on the results of analyzing the accumulated images of the extracted surveillance target.

[0056] The image search unit 34 can be implemented by various known information processing devices that implement algorithms for quickly searching for specific features (e.g., color, shape, or human movement) from image data. A specific example of such an information processing device is the information processing device disclosed by the present inventor in Japanese Patent No. 6593949. This information processing device has a category designation unit, an object (monitoring target) extraction unit, an object (monitoring target) analysis unit, and an analysis unit, and has a function of extracting, from recorded images (accumulated images), analysis target objects (monitoring targets) that belong to an analysis target category designated by the category designation unit.

[0057] In the information processing device according to the above-mentioned patented invention, the extraction of an object (monitoring image) by the object (monitoring target) extraction unit is performed at high speed by a machine learning type image recognition means having a neural network, a so-called deep learning type image recognition means.

[0058] Although the algorithm of the image recognition processing means for extracting objects (monitoring targets) from recorded images (accumulated images) is not particularly limited, "You only look once (YOLO)" can be preferably used. For example, by using "You only look once (YOLO)" as the image recognition means for extracting and identifying objects (monitoring targets) in the object (monitoring target) extraction unit, it is possible to simultaneously and individually extract approximately 1,000 types of analysis target objects (monitoring targets). By utilizing this function, useful objects required by the security monitoring system 100 for more optimal monitoring can be quickly and accurately extracted from accumulated images accumulated within a certain period of time in the past.

[0059] It is also effective to provide the image search unit 34 with a function for tagging people and objects to be monitored. This allows tagged monitored objects to be searched for and detected more quickly, dramatically improving the detection accuracy and speed of the AI ​​target detection means. Furthermore, tagging also allows image data to be transmitted in some communications by simply exchanging tag information, significantly reducing the increase in communication volume over the network in the security monitoring system 100 and the processing load on the server side.

[0060] Furthermore, in the information processing device according to the above-mentioned patented invention, it is also effective to equip the object (surveillance target) extraction unit with a function that can individually recognize the gender and age of people using a face recognition function, which makes it possible to specify the analysis target category as an attribute such as "women in their 30s" when searching stored images.

[0061] [Monitoring necessity information update section] The monitoring necessity information update unit 35 has a monitoring necessity information update means that, when the user 5 inputs a judgment result regarding whether or not to continue monitoring an object requiring confirmation that has been output as warning information by the monitoring information notification unit 33, dynamically updates the content of the monitoring necessity information stored in the monitoring necessity information storage unit 32 to be consistent with the above judgment result of the user 5.

[0062] (Monitoring necessity information update means) The monitoring necessity information updating means constituting the monitoring necessity information updating unit 35 can be configured by various kinds of arithmetic processing means having the function of dynamically updating the contents of the monitoring necessity information so that it matches the above-mentioned judgment result of the user 5. Specific details of this updating process will be described later in the explanation of the embodiment of the "security monitoring method" of the present invention.

[0063] [User device] In the security monitoring system 100, the user 5 receives the monitoring information ("alert information" or "warning information") output by the monitoring information notification unit 33 through the user terminal 4. Therefore, the user terminal 4 must first have a function that enables the user 5 of the security monitoring system 100 to receive the monitoring information ("alert information" or "warning information") output from the monitoring information notification unit 33.

[0064] Additionally, the user 5 also transmits the result of his / her judgment as to whether future monitoring is necessary for the received monitoring information via the user terminal 4 to the monitoring necessity information update unit 35. Therefore, the user terminal 4 must secondly have a function of inputting the user's own judgment as to whether future monitoring is necessary for the received monitoring information, and transmitting the judgment result to the monitoring necessity information update unit 35. In detail, it is preferable that the user 5 be able to input the judgment result of "monitoring necessary" or "monitoring not necessary" with one click on the monitoring information notification screen.

[0065] As long as the device has the first and second functions, the user 5 of the security monitoring system 100 can use, as the user terminal 4, various types of information processing terminals that can appropriately display the above monitoring information in recognizable images, text, audio, etc., such as a small portable information processing terminal such as an easily portable smartphone, or a personal computer with a monitor, by installing a dedicated application for performing the first and second functions. In other words, the user 5 can use the security monitoring system 100 and enjoy its crime prevention effects by using various information processing terminals that he or she has previously owned or been loaned as the user terminal 4.

[0066] <Security monitoring method (operation of security monitoring system)> Fig. 2 is a flowchart showing the flow of the security monitoring method of the present invention (hereinafter, also simply referred to as the "security monitoring method"). As shown in Fig. 2, the "security monitoring method" is a process that cyclically performs a photographing step st1 (acquiring a monitoring image), a recording step st2 (recording and saving the monitoring image), a stored image search step st3 (adjusting the suspiciousness judgment criteria using the stored image data), a confirmation target detection step st4 (automatic detection of a monitoring target (a monitoring target to be confirmed) by artificial intelligence (AI)), a monitoring necessity determination step st5 (determining whether the confirmation target and the monitoring target are the same), a monitoring information notification step st6 (notifying monitoring information (warning information or alert information)), a monitoring necessity information update step st7 (manual update of the monitoring necessity information by the user of the security monitoring system), and a monitoring end step st8.

[0067] The "security monitoring method" according to the above embodiment (hereinafter referred to as the "first embodiment") can be implemented using the security monitoring device 100 of the present invention, which is equipped with a photographing unit 1, a recording unit 2, an image analysis unit 31, a monitoring necessity information storage unit 32, a monitoring information notification unit 33, an image search unit 34, and a monitoring necessity information update unit 35. First, the "security monitoring method" according to this "first embodiment" will be described in detail below.

[0068] [Shooting steps] In the photographing step st1, the photographing unit 1 photographs the area to be monitored and acquires monitoring images in real time or periodically. The monitoring images photographed in the photographing step st1 are transmitted as digital image data to each unit that performs subsequent processing.

[0069] [Recording Steps] In the recording step st2, the surveillance images acquired in the photographing step st1 are transmitted in real time to the recording unit 2, where they are recorded and stored as digital data. In this specification, the surveillance images stored in the recording unit 2 in this manner are referred to as "stored images." It is preferable to reduce the data volume of the stored images by using a compression algorithm (H.264, H.265) during recording in the recording step st2. It is also preferable to store the image data of the stored images in various storages on a cloud server or in various stand-alone storage devices connected to the network, and manage them in a data structure that allows for rapid access in subsequent steps.

[0070] Furthermore, in the recording step st2, it is preferable to store the "accumulated images" in the recording unit 2 with various metadata (for example, "date and time of shooting (timestamp)," "location of shooting (camera ID and coordinate information)," "characteristics of the subject (attribute information of the detected person or object)," etc.) added. By automatically storing past footage from a certain period of time in this data format in the recording step st2, it is possible to quickly search and extract the required image data from the accumulated images based on a specific date and time or characteristics, such as specifying "the person who entered the garden last night."

[0071] Furthermore, it is preferable that the recording of the surveillance images in the recording step st2 is performed continuously 24 hours a day, 365 days a year, but depending on the purpose of the surveillance, it is also possible to perform the recording in a manner that starts only when an abnormality or a sign of an abnormality is detected in the monitored area, for example, through image analysis of the surveillance images by artificial intelligence (AI).

[0072] [Stored Image Search Step] In the stored image search step st3, the image search unit 34 searches the stored images for images of specific monitoring targets that may contribute to determining the degree of suspiciousness in extracting targets to be identified, based on specific search criteria (e.g., time, location, target characteristics). This search allows image data of the monitoring targets to be extracted as information including metadata related to behavioral history and movement routes within the monitoring target area. Specifically, for example, image data that meets search criteria such as "person wearing a red hat" and "vehicle number XYZ123" can be extracted and sent to each unit that will perform subsequent processing.

[0073] [Check target detection step] In step st4 of detecting an object requiring confirmation, the image analysis unit 31 performs image analysis of the surveillance image using artificial intelligence (AI), and as a result, any surveillance object that the artificial intelligence (AI) determines to have a certain level of suspiciousness or above (above a predetermined threshold value set in advance) is detected as an object requiring confirmation.

[0074] Detection of targets requiring investigation is performed by using an artificial intelligence (AI) image analysis function to extract features of targets (people, vehicles, etc.) in surveillance images, recognizing people and objects, and analyzing the movements and behavior of the recognized targets (people, vehicles, etc.) over time to calculate a suspiciousness score. Targets determined to have a suspiciousness score above a certain level are then labeled and detected as "targets requiring investigation," and transmitted to the respective units that perform subsequent processing. In addition, in step st4 of detecting targets requiring investigation, the image analysis unit 31 works in cooperation with the multiple cameras that make up the imaging unit 1 to comprehensively analyze information from the entire area to be monitored, thereby analyzing the associations and interrelationships between multiple monitored targets and generating more accurate alert and warning information.

[0075] In addition, in the confirmation target detection step st4, the criteria for determining the suspiciousness of the monitoring target are automatically adjusted as appropriate according to the results of analyzing the stored images. This adjustment of the suspiciousness criteria is performed by collecting data on the frequency of past detection, the location of detection, and characteristic movements of each monitoring target detected by the confirmation target detection means by analyzing the stored images, and adjusting the process of appropriately changing the suspiciousness criteria for each monitoring target based on the obtained new data.

[0076] As a specific example of an embodiment of adjusting the suspiciousness determination criteria, the processing shown in Table 1 below can be exemplified. For example, as shown in Table 1, if "for a certain vehicle in a surveillance image, the same vehicle is also searched for in stored images, and it is determined as an analysis result that the vehicle appears in the same location with a certain frequency or more," then "the weighting of the suspiciousness score calculation model by artificial intelligence (AI) can be changed so that the suspiciousness score (a numerical value indicating the level of suspiciousness) for the vehicle is calculated to be higher. Alternatively, this adjustment may be made by lowering the threshold for determining suspiciousness so that a judgment result of suspiciousness is more likely to be output. In any case, by automatically performing dynamic adjustments using past surveillance data (stored images) in this way, the suspiciousness determination criteria can be continuously adapted to environmental changes specific to the actual surveillance area and behavior patterns of specific targets, without imposing additional workloads on the system administrator or user, such as manually searching past data.

[0077] [Table 1]

[0078] [Monitoring necessity determination step] In the monitoring necessity determination step st5, the monitoring necessity determination means implemented in the image analysis unit 31 compares the image data of the "target to be confirmed" detected in the target to be confirmed detection step st4 with the image data of the "target to be monitored" stored in the monitoring necessity information storage unit 32, or the image data of the "target to be confirmed" with the image data of the "target not to be monitored", and automatically determines the identity (or non-identity) of the two.

[0079] [Monitoring information notification step] In the monitoring information notification step st6, based on the determination result in the monitoring necessity determination step st5, the monitoring information notification unit 33 outputs "alert information" or "warning information" and transmits it in real time to the user 5. The "alert information" and "warning information" include detailed information such as an image of the monitored object, a suspiciousness determination and the reason for the suspiciousness determination, the location of detection, and the time.

[0080] In addition, "alert information" is applied to targets with high urgency, and is notified to the user 5 via the user terminal 4 by an audio alarm or highlighted display. On the other hand, "warning information" is applied to targets with a relatively low risk, and the user 5 is notified of detailed information about the detected target to be monitored so that the user 5 can quickly determine whether or not monitoring is necessary, and the system prompts the user 5 to confirm whether or not monitoring is necessary.

[0081] Specifically, as shown in Table 2 below, if the object to be checked detected in monitoring necessity determination step st5 is determined to be identical to the "object to be monitored," "alert information" about the object to be checked is output, and if the object to be checked is determined to be not identical to the "object to be monitored," "warning information" about the object to be checked is output.

[0082] Also, as shown in Table 2, if the subject requiring confirmation is determined to be the same as a "subject not requiring monitoring," it is possible to prevent any information from being output. As a result, even if an object is detected as a person or object that artificial intelligence (AI) has initially determined to be suspicious, if the "user" has determined that the object does not require monitoring based on his or her own criteria, the "user" can quickly exclude the object from the list of subjects requiring monitoring with an easy-to-perform operation without relying on an excessive burden on the "system administrator," thereby enabling system optimization through dynamic updating of the content of monitoring necessity information to proceed more efficiently and without omissions.

[0083] [Table 2]

[0084] (Monitoring necessity information update step) In monitoring necessity information update step st7, the monitoring necessity information update unit 35 dynamically updates the content of the "monitoring target" stored in the monitoring necessity information storage unit 32 based on the judgment result of the user 5. For example, if the user 5 judges that a target to be confirmed (e.g., a suspicious person) output as caution information to the user 5 by the monitoring information notification unit 33 is "an acquaintance of the user 5 and further monitoring is unnecessary" and transmits an instruction to that effect from the user terminal 5 to the processing unit 3 constituting the security monitoring system 100 via a telecommunications line, the monitoring necessity information update unit 35, upon receiving the instruction, executes a process of registering the suspicious person in a list of targets not requiring monitoring in the monitoring necessity information. Similarly, if the user 5 determines that a target requiring confirmation (e.g., a suspicious vehicle) output as warning information is a suspicious vehicle and that further monitoring is necessary, and transmits an instruction to that effect from the user terminal 5 to the processing unit 3 constituting the security monitoring system 100 via a telecommunications line, the monitoring necessity information update unit 35, upon receiving the instruction, executes a process of registering the suspicious vehicle in the list of targets requiring monitoring in the monitoring necessity information. This makes it possible to constantly and continuously make detailed judgments on the level of suspiciousness that are suited to the specific circumstances of the area to be monitored.

[0085] The "security monitoring method" of the present invention can also be a process (second embodiment) in which, among the above steps, only the photographing step st1, the confirmation target detection step st4, the monitoring necessity determination step st5, the monitoring information notification step st6, the monitoring necessity information update step st7, and the monitoring termination step st8 are cyclically performed. The "security monitoring method" according to this second embodiment can also be implemented using the "security monitoring device" of the present invention. In this case, of the above-mentioned components, the recording unit 2 and the image search unit 34 are not essential, and the "security monitoring method" of the present invention can be implemented by a "security monitoring device" of the embodiment that has the other components described above.

[0086] Alternatively, the "security monitoring method" of the present invention can be a process (third embodiment) in which only the photographing step st1, the recording step st2, the stored image search step st3, the confirmation target detection step st4, the monitoring information notification step st6, and the monitoring termination step st8 are cyclically performed among the above steps. The "security monitoring method" according to this third embodiment can also be implemented using the "security monitoring device" of the present invention. In this case, among the above-mentioned components, the monitoring necessity information storage unit 32 and the monitoring necessity information update unit 35 are not essential, and the "security monitoring method" of the present invention can be implemented by the "security monitoring device" of the embodiment having the other components described above. [Example]

[0087] A more specific "embodiment" of the "security monitoring system" of the present invention will be exemplified below.

[0088] [Example 1] <Example of a "security monitoring system" in a high-end residential area> [System Overview] (Installation location) A private residence in an upscale residential area with a large lot, multiple entrances, gardens, and garages. (Purpose of use of the system) Ensuring the safety of residents, detecting suspicious people and vehicles, and monitoring mail and deliveries. [Monitoring flow] (camera installation) High-resolution IP cameras are installed in the garden and in front of the garage as camera unit 1. The cameras operate 24 hours a day, capturing surveillance images in real time (photography step st1). (Recording) The monitoring image is stored as a stored image in the recording unit 2, and metadata (shooting time, camera ID, target characteristics) is automatically added (recording step st2). (Trigger for anomaly detection) A vehicle parked in front of the garage, and Camera Unit 1 (high-resolution IP camera) detected that the vehicle had been parked there for more than three minutes. (Setting search conditions) The system searches for the history of the vehicle in question from stored images using conditions such as the vehicle color, partial license plate information, and parking location. (Search results) Footage from the past week is searched, and it is determined that the vehicle in question has been parked in the same location several times (stored image search step st3). (Analysis process) The searched stored images are reanalyzed using AI to detect a history of the vehicle appearing multiple times during nighttime no-parking hours, and the vehicle's behavioral patterns of loitering around the garage are confirmed. (Updated detection of suspected targets and criteria for determining suspiciousness) The suspiciousness score increases based on past behavioral history. The AI ​​determines that the person is a "monitoring target" (target requiring investigation) with a certain level of suspiciousness (target requiring investigation detection step st4). (Determination of necessity of monitoring) For the object requiring monitoring, a check is made against the monitoring necessity information to determine whether or not a notification to the user (resident of the managed house) is required (monitoring necessity determination step st5). (Notification to users) The monitoring information is sent to the user's smartphone (monitoring information notification step st6). An example of the notification content is, "There is a vehicle parked in front of the garage for a long time. It has a history of similar behavior over the past week. Please be on the lookout for this suspicious vehicle." A thumbnail image and a link to play the video are attached to the notification, allowing the user 5 to immediately check the situation. (Update of monitoring information) User 5 checks the notification content and selects "Add as subject to monitoring" or "No monitoring required" with one click on their smartphone. This allows the system to dynamically update the monitoring subject list and reflect this in future decisions (monitoring necessity information update step st7). For example, if the detected subject to be checked is a family member or acquaintance, they can be registered as an authorized visitor, and future notifications will no longer be necessary.

[0089] In the embodiment of Example 1, it is possible to improve the safety of private residences and prevent crimes by detecting suspicious individuals early. Furthermore, by utilizing past data and continuously optimizing the criteria for suspicious behavior, automatic analysis and notification by AI is performed in a cyclical manner, which reduces false alarms and reduces the burden on users (residents) to respond, while strengthening the security system.

[0090] [Example 2] <Example of "security monitoring system" operation in a parking lot> [System Overview] (Installation location) Parking lot of a large commercial facility. Surveillance cameras are installed to cover parking spaces, vehicle entrances and exits, and pedestrian walkways. Entrance gate camera: Records vehicle license plates and models. Parking space camera: Monitors vehicle parking status and records vehicle movement and movement. Pedestrian walkway cameras: Monitor pedestrian movement and detect suspicious behavior. (Purpose of use of the system) Ensuring safety in parking lots, detecting suspicious vehicles and behavior, and monitoring abandoned items and parking violations. [Monitoring flow] (Recording) The surveillance camera records images of vehicles and pedestrians (shooting step st1) and stores them in the recording unit 2. Characteristic information such as the vehicle's license plate, color, shape, and parking location is recorded as metadata. The surveillance image is saved as an accumulated image in the recording unit 2, and metadata (shooting time, camera ID, target characteristics) is automatically added (recording step st2). (Trigger for anomaly detection) If a vehicle stops outside a parking space or if a pedestrian behaves suspiciously, this is detected as suspicious behavior. (Setting search conditions) Enter search criteria for vehicles or people (e.g., specific license plates, specific time periods). (Search results) High-speed search of video footage from the past week and listing of relevant footage. Search results include timestamps and vehicle location information (stored image search step st3). (Analysis process) The system analyzes the retrieved stored images and evaluates past behavioral patterns and abnormalities. For example, if a vehicle has been illegally parked in the past, the system will increase its suspiciousness score. (Updated detection of suspected targets and criteria for determining suspiciousness) The suspiciousness score increases based on past behavioral history. The AI ​​determines that the person is a "monitoring target" (target requiring confirmation) with a certain level of suspiciousness (target requiring confirmation detection step st4). (Determination of necessity of monitoring) For the object requiring monitoring, a check is made against the monitoring necessity information to determine whether or not a notification to the user (parking lot manager) is required (monitoring necessity determination step st5). (Notification to users) The monitoring information is notified to the user's smartphone (monitoring information notification step st6). At the time of notification, a thumbnail image and a link to play the video are attached, so that the user 5 can immediately check the situation. (Update of monitoring information) User 5 checks the notification content and selects "Add as subject to monitoring" or "No monitoring required" with one click on the smartphone at hand. This allows the system to dynamically update the monitoring subject list and reflect it in future judgments (monitoring necessity information update step st7).

[0091] In the embodiment of Example 2, it is possible to improve safety in parking lots and prevent accidents and crimes by detecting illegal parking. In addition, by utilizing past data and continuously optimizing the criteria for suspicious behavior, automatic analysis and notification by AI are performed cyclically, which can significantly reduce the workload of users (monitoring personnel). [Explanation of symbols]

[0092] 1. Filming Department 2 Recording section 3. Processing unit 31 Image analysis unit 32 Monitoring necessity information storage unit 33 Monitoring information notification section 34 Image Search Section 35 Monitoring necessity information update section 4. User terminal 5. User (User of security monitoring system) 6. Target of surveillance (user's home) 7 Suspicious Person 8 Suspicious Vehicles 100 Security Monitoring System st1 Shooting step st2 recording step st3 Stored image search step st4 Step to detect objects requiring confirmation st5 Monitoring necessity determination step st6 Monitoring information notification step st7 Monitoring necessity information update step st8 Monitoring end step

Claims

1. A security monitoring system having a monitoring area within a perimeter range of a specific private residence, an imaging unit that acquires a monitoring image of the monitoring target area; a monitoring necessity information storage unit storing monitoring necessity information including a list of monitoring-required targets specific to the monitoring target area; an image analysis unit that detects a target to be confirmed from the surveillance image using an image analysis function of a machine learning type artificial intelligence and determines whether the detected target to be confirmed is the same as the target to be monitored stored in the monitoring necessity information storage unit; a monitoring information notification unit that notifies a user of the security monitoring system of monitoring information; a monitoring necessity information update unit, the monitoring information notification unit has a monitoring information selection output means for outputting, as the monitoring information, warning information that calls for urgent vigilance for the confirmation target to the user when it is determined that the confirmation target is the same as the monitoring target, and for outputting, as the monitoring information, caution information that calls for the user to determine whether or not to continue monitoring the confirmation target to the user when it is determined that the confirmation target is not the same as the monitoring target, the monitoring necessity information update unit includes a monitoring necessity information update means for dynamically updating the content of the monitoring necessity information so as to be consistent with a determination result input by the user as to whether or not the user needs to continue monitoring the confirmation target output as the warning information, a learning means for performing machine learning to enable the artificial intelligence to constantly make judgments suited to the environment and situation specific to the area to be monitored, which differ for each individual user, based on the content of the monitoring necessity information updated by the monitoring necessity information update unit; Security surveillance system.

2. The monitoring necessity information also includes a list of objects that do not need to be monitored that are specific to the monitoring target area, When it is determined that the target to be confirmed is the same as the target not requiring monitoring, the monitoring information notification unit does not output information about the target to be confirmed. The security monitoring system according to claim 1 .

3. 1. A security monitoring method for monitoring a target area within a perimeter range of a specific private residence, comprising: an imaging step in which an imaging unit captures an image of the monitored area to obtain a monitoring image; a confirmation-required object detection step in which an image analysis unit detects a confirmation-required object from the surveillance image using an image analysis function of a machine learning type artificial intelligence; a monitoring necessity determination step in which the image analysis unit determines whether the target to be confirmed is identical to a monitoring target specific to the monitoring target area stored in the security monitoring system; a monitoring information notifying step in which a monitoring information notifying unit notifies a user of the security monitoring system of monitoring information; a monitoring necessity information updating step in which a monitoring necessity information updating unit updates the content of the monitoring necessity information including a list of monitoring-required targets specific to the monitoring target area, In the monitoring information notification step, if it is determined that the confirmation target is the same as the monitoring target, warning information is output as the monitoring information, requesting the user to take urgent precautions regarding the confirmation target, and if it is determined that the confirmation target is not the same as the monitoring target, caution information is output as the monitoring information, requesting the user to determine whether or not to continue monitoring the confirmation target, In the monitoring necessity information updating step, when a result of a determination by the user as to whether or not it is necessary to continue monitoring the confirmation target output as the warning information is input, the content of the monitoring necessity information is dynamically updated to be consistent with the result of the determination, Based on the updated content of the monitoring necessity information in the monitoring necessity information update step, machine learning is performed so that the artificial intelligence can constantly make decisions that are suited to the environment and situation specific to the monitoring target area, which differ for each individual user. Security monitoring methods.

4. The monitoring necessity information also includes a list of objects that do not need to be monitored that are specific to the monitoring target area, In the monitoring information notification step, if it is determined that the target to be confirmed is the same as the target not requiring monitoring, information about the target to be confirmed is not output. The security monitoring method according to claim 3 .

5. 5. A method for monitoring security according to claim 3, wherein the method is performed by a computer. A program for security monitoring.

Images