WO2023113183A1

WO2023113183A1 - Apparatus and method for analyzing children's behavior using edutech technology to improve educational culture for children

Info

Publication number: WO2023113183A1
Application number: PCT/KR2022/015540
Authority: WO
Inventors: 박윤하; 김대수; 김종일; 서우석
Original assignee: 주식회사 우경정보기술
Priority date: 2021-12-14
Filing date: 2022-10-13
Publication date: 2023-06-22
Also published as: KR20230089956A

Abstract

An analysis apparatus is provided. The analysis apparatus may comprise: a collection unit that collects a first image in which a user is captured; a de-identification unit that de-identifies the user appearing in the first image; an analysis unit that analyzes a behavior of the user appearing in the first image; and a storage unit for matching and storing a second image corresponding to the first image de-identified by the de-identification unit and analysis information analyzed by the analysis unit.

Description

Apparatus and method for analyzing children's behavior using edutech technology to improve educational culture for children

The present invention relates to a device and method for analyzing the behavior of a user such as a child and providing various services based on the analysis result.

Safety accidents and childcare accidents caused by insufficient infrastructure in child-related facilities are becoming social problems.

As a result, the need for early identification of children in need of treatment and customized education and care services is increasing.

It is an object of the present invention to provide an analysis device and method that analyzes a user's behavior and provides various services based on the result of the behavioral analysis while minimizing exposure of personal information.

The analysis device of the present invention includes a collection unit for collecting a first image captured by a user; a de-identification unit that de-identifies the user appearing in the first image; an analyzer configured to analyze a behavior of the user appearing in the first image; It may include a storage unit for matching and storing the second image corresponding to the first image de-identified by the de-identification unit and the analysis information analyzed by the analysis unit.

The analysis method of the present invention includes a collection step of collecting a first image captured by a user; a de-identification step of de-identifying the user appearing in the first image; a pre-processing step of selecting a target image in which the user appears from among a plurality of first images output from the collecting step and standardizing only the plurality of target images; Analyzing changes in user areas scraped from consecutive frames of the standardized target image to extract deviant behaviors of the user, and storing analysis information on the scraped user areas and the user's deviant behaviors in frame units in a scrapbook analysis phase; A re-identification step of identifying whether the user analyzed for deviant behavior by the analyzer is a pre-registered first user, and outputting only the deviant behavior analysis information included in the scrapbook of the first user and the scraped user area among a plurality of scrapbooks. ; and a storage step of matching and storing a second image corresponding to the first image de-identified in the de-identification step, the analysis information of the first user output from the re-identification step, and the scraped user area. can

The analysis apparatus and method of the present invention may analyze abnormal behavior of users such as children through image analysis.

In addition, the analysis device may de-identify and store the image that is the basis for determining the abnormal behavior.

The de-identified image and the user's behavior analysis information may be provided to various service means in pairs.

Due to non-identification processing, the user's behavioral analysis information can be utilized by various service means related to accident prevention, safety, education, psychological counseling, and the like.

According to the present invention, a safer educational environment can be created through child behavior analysis and artificial intelligence-based smart environment service. To this end, the present invention can provide a service platform based on child behavior analysis using edutech technology for improving educational culture for children.

1 is a schematic diagram showing the analysis device of the present invention.

2 is a flowchart illustrating the operation of the collection unit.

3 is a flowchart illustrating the operation of the non-identification unit.

4 is a flowchart illustrating the operation of a pre-processing unit.

5 is a flowchart illustrating the operation of the analyzer.

6 is a flowchart illustrating the operation of the re-identification unit.

7 is a flowchart showing the analysis method of the present invention.

8 is a diagram illustrating a computing device according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily carry out the present invention. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

In this specification, redundant descriptions of the same components are omitted.

In addition, in this specification, when a component is referred to as being 'connected' or 'connected' to another component, it may be directly connected or connected to the other component, but another component in the middle It should be understood that may exist. On the other hand, in this specification, when a component is referred to as 'directly connected' or 'directly connected' to another component, it should be understood that no other component exists in the middle.

In addition, terms used in this specification are only used to describe specific embodiments and are not intended to limit the present invention.

Also, in this specification, a singular expression may include a plurality of expressions unless the context clearly indicates otherwise.

In addition, in this specification, terms such as 'include' or 'having' are only intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more It should be understood that the presence or addition of other features, numbers, steps, operations, components, parts, or combinations thereof is not precluded.

Also in this specification, the term 'and/or' includes a combination of a plurality of listed items or any item among a plurality of listed items. In this specification, 'A or B' may include 'A', 'B', or 'both A and B'.

Also, in this specification, detailed descriptions of well-known functions and configurations that may obscure the subject matter of the present invention will be omitted.

1 is a schematic diagram showing an analysis device 100 of the present invention.

The analysis device 100 shown in FIG. 1 includes a collection unit 110, a de-identification unit 120, a pre-processing unit 130, an analysis unit 140, a re-identification unit 150, a storage unit 160, an API A server 170 and a control unit 180 may be included.

The collection unit 110 may collect a first image in which a user is photographed. The first image may include a video.

For example, the collecting unit 110 may include a photographing means 10 such as a camera, a Closed Circuit TV (CCTV), or a smartphone for photographing a user. When the photographing means 10 is provided separately, the collection unit 110 may include various communication modules that communicate with the photographing means 10 by wire or wireless.

The de-identification unit 120 may de-identify a user appearing in the first image. De-identification processing may mean processing (including image processing) so that a user appearing in the first image is not recognized. The first image may include a face, a resident registration number, a phone number, a vehicle number, an address, a company name, and the like, which can identify the user's personal information. The first image containing these personal information cannot be provided to a third party due to moral regulations or laws. According to this, it may be difficult for a user to use various service means involving a third party. When the personal information is deleted by the de-identification unit 120, the user can comfortably provide images corresponding to the conditions for receiving the service without any difficulty.

For example, a user with a violent tendency may wish to receive a psychological counseling service. In this state, if the provision of the first image required to accurately analyze the user's condition is limited, it is difficult to provide realistic counseling through a counselor.

However, according to the non-identification unit 120, a so-called face mosaic-processed image can be provided to the counselor, and the counselor can conduct psychological counseling with the user through this.

The analysis unit 140 may analyze the user's behavior appearing in the first image.

In the storage unit 160 , the second image corresponding to the first image de-identified by the de-identification unit 120 and the analysis information analyzed by the analysis unit 140 may be matched and stored.

2 is a flowchart illustrating the operation of the collection unit 110.

The collection unit 110 may receive a new video corresponding to the first video from the photographing unit 10 (S511). The collecting unit 110 may transmit the collected first image to the de-identifying unit 120 in order to de-identify personal information included in the image (S512). Also, the collecting unit 110 may transmit the collected first image to the pre-processing unit 130 (S513).

When the first image is acquired from the photographing unit 10, the collection unit 110 may set some information included in the first image as identification information or set newly generated identification information to the image. The identification information at this time may not be used to identify the user, but may be used to distinguish the first image from other images.

The collection unit 110 may provide the first image for which identification information is set to the non-identification unit 120 and the analysis unit 140, respectively.

Identification information included in the first image may also be included in the second image output from the non-identifying unit 120 . Identification information included in the first image may also be included in analysis information output from the analyzer 140 .

The storage unit 160 may match the second image and the analysis information by using the identification number included in the second image and the identification information included in the analysis information.

3 is a flowchart illustrating the operation of the non-identification unit 120.

Before storing the collected videos in the storage unit 160, the de-identification unit 120 identifies objects related to personal information, for example, face, resident registration number, phone number, vehicle number, address, area where the company name is displayed, and the object. can be identified.

For example, the de-identification unit 120 may detect an object representing personal identification information included in the first image and damage an area of the object in the first image to be unrecognizable. The de-identified first image may be stored in the storage unit 160 .

The de-identification unit 120 may first receive the first image to be de-identified from the collection unit 110 (S521). At this time, the non-identification unit 120 may read (load) the first image frame by frame (S522).

The de-identification unit 120 may detect an object representing personal identification information (personal information) in units of frames of the first image (S523).

The non-identification unit 120 may mosaic-process the area of the detected object and output a second image corresponding to the mosaic-processed first image (S524). The second image may refer to an image in which a region in which personal identification information is displayed in the first image is damaged or mosaic-processed. In this way, all personal identification information elements included in the first image can be de-identified. When the first image ends (S525), the second image subjected to de-identification processing may be stored in the storage unit 160 (S526).

The pre-processing unit 130 may divide the plurality of first images into a target image and a dummy image.

The target image may include an image in which a user to be analyzed by the analyzer 140 appears among a plurality of first images.

The dummy image may include an image in which a user to be analyzed by the analyzer 140 does not appear among the plurality of first images.

The pre-processing unit 130 may exclude the dummy image from among the target image and the dummy image and provide only the target image to the analysis unit 140 .

Alternatively, the pre-processing unit 130 may divide the single first image into a target section and a dummy section.

The target section may refer to a frame section in which a user to be analyzed by the analyzer 140 appears.

The dummy section may refer to a frame section in which a user to be analyzed by the analyzer 140 does not appear.

The preprocessor 130 may exclude the dummy section and generate a new simplified image using only the target section.

The preprocessor 130 may provide the simplified image to the analyzer 140 instead of the first image.

Meanwhile, the pre-processing unit 130 may perform standardization to convert the format of the first image according to the input format of the analysis unit 140 . For example, the pre-processing unit 130 may convert at least one of the file size, frame per second (FPS), and image format of the first shape according to the input format of the analysis unit 140 . While properties such as size and FPS of the first image vary according to setting values of the photographing unit 10 and collection environments, the form of input data of the artificial intelligence model may be fixed. Accordingly, the properties of the first image need to be modified to match the shape of the input data of the artificial function model. The size and FPS of the first image may be standardized according to the artificial intelligence model loaded in the analysis unit 140 by the pre-processing unit 130 . For example, the first image may be standardized according to the input data form of the analysis model loaded in the analysis unit 140 and the re-identification model provided in the re-identification unit 150 through a pre-processing process.

Preferably, the target image selection and standardization performed by the pre-processing unit 130 are performed together.

4 is a flowchart illustrating the operation of the pre-processing unit 130.

The pre-processing unit 130 may receive the first image from the collecting unit 110 (S531).

The pre-processing unit 130 may select and standardize a target image so that the artificial intelligence model loaded in the analysis unit 140 can analyze the first image. Through image screening and standardization, the processing efficiency of artificial intelligence models can be improved.

A selection module and a standardization module may be provided in the preprocessing unit 130 .

The selection module may select only the target image in which the user appears (S533) from among the plurality of first images output from the collection unit 110 (S532). According to the screening module, unnecessary images are excluded from analysis by the analyzer 140, and the processing efficiency of the entire system can be improved.

The standardization module may standardize only the target image among the plurality of first images (S534). In this case, standardization may include an operation of converting the format of the target image according to the input format of the analyzer 140 .

The standardization module may provide the standardized target image to the analysis unit 140 (S535).

5 is a flowchart illustrating the operation of the analyzer 140.

The analysis unit 140 may receive the first image or simplified image that has passed the pre-processing process from the pre-processing unit 130 (S541).

For example, the analyzer 140 may load the first image in frame units (S542).

The analysis unit 140 may detect and scrap the user area included in each frame (S543).

The analyzer 140 may analyze changes in the user area scraped from consecutive frames, and may extract and analyze the user's abnormal behavior through the analysis of the area change (S544).

The analyzer 140 may match the scraped user area with the extracted abnormal behavior analysis information and store it in a scrapbook (S545). The scrapbook is a set of data generated by the analyzer 140, and may include a plurality of pairs of scraped user areas and abnormal behavior analysis information. The scraped user area may be a cutout of only the user's body part excluding the background in a specific frame. Analysis information of abnormal behavior may include vector information of the user's movement or evaluation results of abnormal behavior analyzed through a machine learning model, for example, classification results such as violence, fall, fall, quarrel, collision, and violence. can

The analyzer 140 may output a specific scrapbook created targeting the specific first image when the analysis of the specific first image is completed or the first image ends (S548). A specific scrapbook may include all scraps and analysis information extracted from a specific first image. The scrapbook may be transferred to the storage unit 160 or the re-identification unit 150 .

The analysis unit 140 may analyze the user's behavior appearing in the first image and detect the user's abnormal behavior through the analysis.

The analysis unit 140 sends a notification message to the control unit 180 or the API server 170 'API server' that manages user-related services when an abnormal behavior, in particular, an emergency abnormal thawing related to a safety accident is detected (S546). can be transmitted (S 546). The notification message may include analysis information and a scraped user area (image). A supervisor or guardian who recognizes the notification message can quickly solve a dangerous situation for the user or his or her surroundings.

6 is a flowchart illustrating the operation of the re-identification unit 150.

The re-identification unit 150 may identify whether the user analyzed by the analysis unit 140 as an abnormal behavior is a pre-registered first user. At this time, the second image and analysis information stored in the storage unit 160 may be limited to those of the first user. The second image and analysis information stored in the storage unit 160 may be used for various services for a specific user. In order to provide a service, it is necessary to specify a user to be provided with the service. Even if a specific user included in the first image shows an abnormal behavior, there is a practical problem in that the service cannot be provided even if the user wants to provide the service because they do not know where they live or who they are. As a result, since analysis of an unspecified user included in the first image has no practical benefit, it may be advantageous to store only the second image and analysis information for the already registered first user in the storage unit 160. .

As described above, the analyzer 140 may store analysis information on the scraped user region and the user's abnormal behavior in frame units of the first image in a scrapbook.

At this time, the re-identification unit 150 may receive the scrapbook from the analysis unit 140 (S551). Then, analysis information and scrap images may be loaded in the scrapbook (S552).

The re-identification unit 150 may use the scraped user area to identify (re-identify) whether a specific user has already been registered (S553).

When a specific user is identified as a pre-registered first user (S554), the re-identification unit 150 stores the abnormal behavior analysis information of the specific user and the scrapped user area (scrap image) in the storage server corresponding to the storage unit 160. ) can be stored (S 555).

The above operations of the re-identification unit 150 may be continuously performed until the scrapbook ends (S556).

The re-identification unit 150 may collect deviant behavior information of the first user through re-identification using a machine-learned re-identification model. Based on the collected information, analysis services such as emotion recognition and behavior patterns through experts and artificial intelligence can be performed. Based on the collected user information, services such as personalized learning and risk prediction can be provided.

A first user may be registered in advance. For example, a feature vector of the first user may be obtained by inputting an image of the first user to a pretrained artificial intelligence model, and a distance between feature vectors may be compared to specify or identify the feature vector of the first user.

The re-identification model has a characteristic of deriving an approximate feature vector even if the shooting environment of the object image is changed. Therefore, it is possible to identify a user by comparing distances between feature vectors of images collected in various shooting environments.

Registration of the first user may be performed by storing a feature vector obtained by inputting the image of the first user into the re-identification model in the user database. It is preferable to extract a feature vector using the front, side, and back images of the first user so that the re-identification performance is maintained even when the direction is changed.

In the process of performing the re-identification unit 150, the scrapbook transmitted from the analysis unit 140 is first received. It reads analysis information and scraped user area (scrap image) from the received scrapbook. Identification is performed by inputting the scrapped user area into the re-identification model and comparing the obtained feature vector with feature vectors of pre-registered first users. When identified as the first user, the user behavior analysis information and the scraped user area may be stored in the storage unit 160 .

The controller 180 may provide functions for managing users such as children and educational facilities. The control unit 180 may support accident response and customized education by providing the collected second video, analysis information, notification message, etc. to educational facilities, parents, and experts. In addition, the control unit 180 may manage and supervise information collected through artificial intelligence.

The control unit 180 may be composed of educational facilities, parents, and experts.

Educational facilities may provide safety and childcare assistance services. Videos collected from multiple sensors are detected and analyzed through artificial intelligence, and the analysis results can be provided to education facility personnel (supervisors, etc.). Notifications may be provided to the person in charge of the educational facility in the event of a dangerous situation, such as a child falling and fighting. In addition, behavioral patterns collected through follow-up observation are provided to the person in charge of educational facilities and can be used as basic data to provide a customized childcare environment.

The controller 180 may provide various types of user information to the guardian. For example, the guardian may check the location information of the child in the educational facility. Through this, it is possible to check whether or not to go to and from school or to prevent various ranges.

The control unit 180 provides experts with abnormal behaviors and data related to children so that they can manage them individually. Individual management systems such as child psychology and psychiatry can be established by providing the collected behavioral and psychological data of children, such as behavioral information, to experts. The quality of personalized education and care can be improved by providing feedback on children's behavior by experts.

An API (Application Programming Interface) server connects the platform and the control unit 180 to provide various services. For example, the API server 170 may provide functions such as first user registration and notification messages to educational facilities, parents, and experts. API services for real-time accident prevention and response and customized training can be provided.

The main functions of the API server 170 are four and are as follows.

1. Re-identified User Management

A function of managing a re-identified user corresponding to the first user is provided. Provides re-identification user registration, inquiry and deletion functions. You can register and track and manage target children you want to observe.

2. Inquiry of abnormal behavior list

Records of detected abnormal behaviors can be queried. You can receive abnormal behavior information such as abnormal behavior time, behavior, and snapshots for each registered user. It is possible to track and observe the child's abnormal behavior, so it is possible to support personalized care learning.

3. Video Management

Recorded video can be managed. An education facility manager (supervisor) may add, download, or delete images collected from the photographing unit 10 .

4. Send notifications

Notifications can be sent when risky abnormal behavior is detected. When an abnormal behavior is detected by the analysis unit 140, a notification message is delivered to the educational facility. Send notifications to the person in charge of the educational facility (supervisor) to assist in immediate response to emergency situations.

The storage unit 160 may include a media database and a user database.

The first image or the second image collected through the collection unit 110 may be stored in the media database. Preferably, only the second image is stored in the media database.

Analysis information of the abnormal behavior detected by the analyzer 140 may be stored in the user database.

7 is a flowchart showing the analysis method of the present invention.

The analysis method of FIG. 7 may be performed by the analysis device 100 shown in FIG. 1 .

The analysis method includes a collection step (S 510), a non-identification step (S 520), a preprocessing step (S 530), an analysis step (S 540), a re-identification step (S 550), a storage step (S 560), a control step, It may include service steps.

In the collecting step (S510), a first image captured by a user may be collected. The collecting step (S510) may be performed by the collecting unit 110.

* In the de-identification step (S 520), a user appearing in the first image may be de-identified. The non-identification step (S520) may be performed by the non-identification unit 120.

In the preprocessing step (S530), a target image in which the user appears is selected from among the plurality of first images output from the collection step (S520), and standardization may be performed only on the plurality of target images. The pre-processing step (S530) may be performed by the pre-processing unit 130.

In the analysis step (S540), the deviant behavior of the user is extracted by analyzing the change in the clipped user area from the continuous frames of the standardized target image, and the analysis information on the scrapped user area and the user's deviant behavior on a frame-by-frame basis is stored in the scrapbook. can be stored in The analysis step (S540) may be performed by the analyzer 140.

In the re-identification step (S550), it is identified whether the user analyzed as deviant behavior by the analyzer 140 is a pre-registered first user, and deviant behavior analysis information and scraps included in the scrapbook of the first user among a plurality of scrapbooks. Only the user area that has been selected can be output. The re-identification step (S550) may be performed by the re-identification unit 150.

In the storage step (S560), the second image corresponding to the first image de-identified in the de-identification step, the analysis information of the first user output from the re-identification step, and the scraped user area may be matched and stored. The storage step (S560) may be performed by the storage unit 160.

In the control step, when an emergency abnormal behavior (emergency situation, emergency situation) is detected in the analysis step, a notification message may be transmitted to a supervisor's terminal supervising the user or a guardian's terminal of the user. The control step may be performed by the control unit 180 .

In the service step, the analysis information of the first user stored in the storage unit 160 and the scrapped user area may be provided to the service means through the storage step. The service step may be performed by the API server 170.

At least one of the de-identification step, the pre-processing step, the analysis step, the re-identification step, the control step, and the service step may be implemented through various artificial intelligence models generated through machine learning.

8 is a diagram illustrating a computing device according to an embodiment of the present invention. The computing device TN100 of FIG. 8 may be a device described in this specification (eg, the analysis device 100, etc.).

In the embodiment of FIG. 8 , the computing device TN100 may include at least one processor TN110, a transceiver TN120, and a memory TN130. In addition, the computing device TN100 may further include a storage device TN140, an input interface device TN150, and an output interface device TN160. Elements included in the computing device TN100 may communicate with each other by being connected by a bus TN170.

The processor TN110 may execute program commands stored in at least one of the memory TN130 and the storage device TN140. The processor TN110 may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which methods according to embodiments of the present invention are performed. Processor TN110 may be configured to implement procedures, functions, methods, and the like described in relation to embodiments of the present invention. The processor TN110 may control each component of the computing device TN100.

Each of the memory TN130 and the storage device TN140 may store various information related to the operation of the processor TN110. Each of the memory TN130 and the storage device TN140 may include at least one of a volatile storage medium and a non-volatile storage medium. For example, the memory TN130 may include at least one of read only memory (ROM) and random access memory (RAM).

The transmitting/receiving device TN120 may transmit or receive a wired signal or a wireless signal. The transmitting/receiving device TN120 may perform communication by being connected to a network.

Meanwhile, the embodiments of the present invention are not implemented only through the devices and/or methods described so far, and may be implemented through a program that realizes functions corresponding to the configuration of the embodiments of the present invention or a recording medium in which the program is recorded. And, such implementation can be easily implemented by those skilled in the art from the description of the above-described embodiment.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concept of the present invention defined in the following claims are also provided. belong to the scope of the invention.

Claims

a collection unit that collects a first image captured by a user;

a de-identification unit that de-identifies the user appearing in the first image;

an analyzer configured to analyze a behavior of the user appearing in the first image;

a storage unit for matching and storing a second image corresponding to the first image de-identified by the de-identification unit and analysis information analyzed by the analysis unit;

Analysis device comprising a.
According to claim 1,

When the first image is obtained from the photographing means, the collection unit sets some information included in the first image as identification information or sets newly generated identification information to the image;

The collection unit provides the first image for which the identification information is set to the non-identification unit and the analysis unit, respectively;

The identification information included in the first image is also included in the second image output from the non-identification unit,

Identification information included in the first image is also included in analysis information output from the analyzer,

Wherein the storage unit matches the second image and the analysis information using an identification number included in the second image and identification information included in the analysis information.
According to claim 1,

The non-identification unit detects an object representing personal identification information included in the first image,

Wherein the non-identification unit damages the region of the object in the first image to be unrecognizable.
According to claim 1,

The de-identification unit detects an object representing personal identification information in units of frames of the first image,

The non-identification unit mosaic-processes the region of the detected object, and outputs the second image corresponding to the mosaic-processed first image.
According to claim 1,

A pre-processing unit is provided to divide the plurality of first images into target images and dummy images;

The target image is an image in which a user to be analyzed by the analysis unit appears,

The dummy image is an image in which a user to be analyzed by the analyzer does not appear,

wherein the pre-processing unit excludes the dummy image from among the target image and the dummy image, and provides only the target image to the analyzing unit.
According to claim 1,

A pre-processing unit is provided to divide a single first image into a target section and a dummy section;

The target section is a frame section in which a user to be analyzed by the analyzer appears,

The dummy section is a frame section in which a user to be analyzed by the analyzer does not appear,

The pre-processor excludes the dummy section and generates a new simplified image using only the target section;

Wherein the pre-processing unit provides the simplified image to the analysis unit instead of the first image.
According to claim 1,

A pre-processing unit is provided to perform standardization of converting the format of the first image according to the input format of the analysis unit;

The pre-processing unit converts at least one of a file size, a frame per second (FPS), and a format of the first shape according to an input format of the analysis unit.
According to claim 1,

Selection modules and standardization modules are prepared,

The screening module selects only a target image in which the user appears from among a plurality of first images output from the collection unit;

The standardization module performs standardization on only the target image among the plurality of first images;

The standardization includes an operation of converting a format of the target image according to an input format of the analysis unit,

The normalization module provides the analysis unit with the standardized target image.
According to claim 1,

The analyzer detects the user's abnormal behavior through analysis of the behavior;

The analyzer transmits a notification message to a control unit that manages a service related to the user when the abnormal behavior is detected by the analyzer.
According to claim 1,

The analysis unit loads the first image frame by frame,

The analysis unit scraps the user area included in each frame,

The analysis unit analyzes changes in the user area scraped from consecutive frames;

The analysis unit extracts and analyzes the abnormal behavior of the user through analysis of the change in the area.
According to claim 10,

The analyzer matches the scraped user area with the extracted abnormal behavior analysis information and stores it in a scrapbook;

The analyzer outputs a specific scrapbook generated for the specific first image when the analysis unit completes the analysis of the specific first image.
According to claim 1,

A re-identification unit is provided to identify whether the user analyzed as abnormal behavior by the analysis unit is a pre-registered first user;

The second image and the analysis information stored in the storage unit are limited to those of the first user.
According to claim 12,

The analysis unit stores analysis information on a user region scraped in frame units of the first image and an abnormal behavior of the user in a scrapbook;

The re-identification unit identifies whether a specific user has already been registered using the scraped user area;

Wherein the re-identification unit stores the abnormal behavior analysis information of the specific user and the scraped user area in the storage unit when the specific user is identified as the first user.
In the analysis method performed by the analysis device,

A collection step of collecting a first image captured by a user;

a de-identification step of de-identifying the user appearing in the first image;

a pre-processing step of selecting a target image in which the user appears from among the plurality of first images output from the collecting step and standardizing only the plurality of target images;

Analyzing changes in user areas scraped from consecutive frames of the standardized target image to extract deviant behaviors of the user, and storing analysis information on the scraped user areas and the user's deviant behaviors in frame units in a scrapbook analysis phase;

A re-identification step of identifying whether the user analyzed for deviant behavior by the analyzer is a pre-registered first user, and outputting only the deviant behavior analysis information included in the scrapbook of the first user and the scraped user area among a plurality of scrapbooks. ;

a storage step of matching and storing a second image corresponding to the first image de-identified in the de-identification step, the analysis information of the first user output from the re-identification step, and the scraped user area;

Analysis method comprising a.
According to claim 14,

a control step of transmitting a notification message to a terminal of a supervisor supervising the user or a terminal of a guardian of the user when an emergency or abnormal behavior is detected in the analysis step;

and a service step of providing the analysis information of the first user and the scraped user area stored in a storage unit through the storing step to a service means.