KR102265951B1 - Cctv sorting control system using deep learning - Google Patents

Abstract

The present invention relates to a CCTV screening control system using deep learning. The task to be solved is to enable more efficient control with fewer people by custom setting the detection elements of various CCTV images and detecting each using deep learning. As an example, disclosed is a CCTV screening control system using deep learning, comprising: CCTV detection option setting unit for setting a detection option including at least one of a control object to be detected in the CCTV image, a specific action for the control object or an object related to the control object, detection date and time, and detection location; a CCTV selection detection unit for selectively detecting a CCTV image according to a detection option set through the CCTV detection option setting unit among a plurality of CCTV images; and CCTV selection notification unit for notifying the detection result of the CCTV image through the CCTV selection detection unit.

Description

CCTV screening control system using deep learning {CCTV SORTING CONTROL SYSTEM USING DEEP LEARNING}

An embodiment of the present invention relates to a CCTV screening control system utilizing deep learning.

In general, the control method in the CCTV integrated control center is to prevent various crimes and illegal activities by having a professional manpower to monitor 48 CCTV images recommended by the Ministry of Public Administration and Security per person in real time.

At this time, the number of CCTVs installed in a specific area is about 13,000, and about 350 control personnel are monitoring 24 hours in 2-3 shifts. The number of CCTV images to be responsible for each person is 111.4, which is more than double the above recommendations. It is a number.

Therefore, the CCTV video control method, which relies on the existing manpower, decreases control efficiency and concentration of video control personnel due to the rapid increase in CCTV, and inevitably causes a huge financial burden. To solve this, improvement of the CCTV video control method is needed. it is necessary.

In addition, it is urgent to secure screening control technology that can effectively support control personnel work by accurately detecting and selecting events and responding to the increasing number of CCTVs in public institutions every year.

For this purpose, intelligent CCTV was developed, but the early intelligent CCTV only implemented simple intelligent functions through beginner-level image change using simple image processing rather than using complex algorithms, and focused on increasing the added value of CCTV rather than actual use. this was matched

In addition, recently released products are limited to the identification of simple intelligent video events (specific objects such as people and cars), so there is a difficulty in control that can select various events suitable for specific and complex situations.

Registered Patent Publication No. 10-2144531 (Registration date: August 07, 2020)

An embodiment of the present invention provides a CCTV screening control system using deep learning that enables high-efficiency control with fewer people by custom setting the detection elements of various CCTV images and detecting each using deep learning. do.

The CCTV selection control system using deep learning according to an embodiment of the present invention is a control object to be detected in a CCTV image, a specific action/action for the control object or an object related to the control object, the detection date and time, and the detection position CCTV detection option setting unit for setting a detection option including at least one; a CCTV selection detection unit for selectively detecting a CCTV image according to a detection option set through the CCTV detection option setting unit among a plurality of CCTV images; and a CCTV selection notification unit for notifying the detection result of the CCTV image through the CCTV selection detection unit.

In addition, the CCTV detection option setting unit, a first detection option setting unit for selecting and setting a first detection option including a control object to be detected in the CCTV image; a second detection option setting unit for selecting and setting a second detection option including a specific action or action for the control object of the first detection option; a third detection option setting unit for selecting and setting a third detection option including at least one of a detection date and time; and a fourth detection option setting unit for selecting and setting a fourth detection option including the detection position.

In addition, the first detection option setting unit, in the case of an upper category detection option including at least one control object among humans, animals, vehicles, and objects, and an upper category detection option for humans, gender, face, race, age, and clothing , body and belongings, select and set a subcategory detection option including at least one characteristic, but in the case of a subcategory detection option for race, select and set a detailed detection option for skin color, and set a subcategory detection option for clothes In the case of , the detailed detection options for clothing type and clothing color are selected and set, and in the case of the sub-category detection option for belongings, the detailed detection options for the type and color of belongings are selected and set, and the upper category detection option for animals In the case of , select and set a sub-category detection option that includes at least one characteristic of animal type, animal size, and animal color. Select and set a sub-category detection option including at least one of set, and color-related detection options may provide color codes.

In addition, the second detection option setting unit may be set by selecting and setting at least one second detection option among intrusion, wandering, escaping, stopping, falling, staggering actions or actions of the control object, or objects related to the control object.

Also, the fourth detection option setting unit may receive and set location information of at least one of a region, a region, and a range for the region or region as the fourth detection option.

In addition, when the CCTV image according to the detection option set through the CCTV selection detection unit is detected, the CCTV selection notification unit may display the detected CCTV image in a pop-up form to notify the detection result of the CCTV image.

In addition, the CCTV detection option setting unit, when the control object cannot be specified, receives the montage information on the control object through at least one of a drawing tool and a photographed image scan to receive a first detection option for setting the first detection option It may further include an exception setting unit.

In addition, when the CCTV detection option setting unit sets at least two groups of the first detection option, the second detection option, the third detection option and the fourth detection option at the same time, the same detection option is the same condition Detection for setting a detection option combination that combines the first detection option, the second detection option, the third detection option, and the fourth detection option for a first condition and a second condition that is different between the first detection option and the same detection option It may further include an option combination setting unit.

According to the present invention, it is possible to provide a CCTV screening control system using deep learning that enables high-efficiency control with fewer people by custom setting the detection elements of various CCTV images and detecting each using deep learning. have.

1 is a schematic diagram showing a working environment to which a CCTV screening control system using deep learning according to an embodiment of the present invention is applied.
2 is a block diagram showing the overall configuration of a CCTV screening control system utilizing deep learning according to an embodiment of the present invention.
3 is a block diagram showing a detailed configuration of a CCTV detection option setting unit according to an embodiment of the present invention.
4 is a view showing an example of the execution screen of the first and second detection option setting unit according to an embodiment of the present invention.
5 is a view showing an example of the execution screen of the third detection option setting unit according to an embodiment of the present invention.
6 is a diagram illustrating an example of an execution screen of a fourth detection option setting unit according to an embodiment of the present invention.
7 is a diagram illustrating an operation of a CCTV selection notification unit according to an embodiment of the present invention.

Terms used in this specification will be briefly described, and the present invention will be described in detail.

The terms used in the present invention have been selected as currently widely used general terms as possible while considering the functions in the present invention, but these may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technology, and the like. In addition, in a specific case, there is a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the corresponding invention. Therefore, the term used in the present invention should be defined based on the meaning of the term and the overall content of the present invention, rather than the name of a simple term.

In the entire specification, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated. In addition, terms such as "...unit" and "module" described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software, or a combination of hardware and software. .

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those of ordinary skill in the art can easily carry out the embodiments of 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.

1 is a schematic diagram showing a working environment to which a CCTV screening control system using deep learning according to an embodiment of the present invention is applied, and FIG. 2 is an overall configuration of a CCTV screening control system using deep learning according to an embodiment of the present invention. 3 is a block diagram showing the detailed configuration of a CCTV detection option setting unit according to an embodiment of the present invention, and FIG. 4 is an execution screen of the first and second detection option setting unit according to an embodiment of the present invention. is a view showing an example of, Figure 5 is a view showing an example of the execution screen of the third detection option setting unit according to the embodiment of the present invention, Figure 6 is the execution screen of the fourth detection option setting unit according to the embodiment of the present invention It is a view showing an example of, Figure 7 is a view showing to explain the operation of the CCTV selection notification unit according to an embodiment of the present invention.

Referring to FIG. 2 , the CCTV selection control system 1000 using deep learning according to an embodiment of the present invention is a CCTV detection option setting unit 100 , a CCTV selection detection unit 200 , and a CCTV selection notification unit 300 . It may include at least one.

The CCTV detection option setting unit 100 includes at least one of a control object to be detected in the CCTV image, a specific action / action for the control object or / and an object related to the control object, detection date and time, and detection location You can select and set the detection option. That is, the CCTV detection option setting unit 100 selects a detection option for who (control object), what (specific action/action), when (date and time), and where (location) by the user (custom) method) can be set.

To this end, the CCTV detection option setting unit 100 includes a first detection option setting unit 110 , a second detection option setting unit 120 , a third detection option setting unit 130 , and a fourth detection option setting unit 140 . may include.

The first detection option setting unit 110 may select and set a first detection option including a control object to be detected in CCTV images received from various CCTV devices. More specifically, the first detection option setting unit 110 may select an upper category detection option and a corresponding lower category detection option as shown in FIG. 4 . Here, the upper category detection option includes at least one control object among humans, animals, vehicles, and objects, and when the upper category detection option for humans is selected, at least one of gender, face, race, age group, clothing, body, and belongings Sub-category detection options including one characteristic can be selected and set.

The first detection option setting unit 110 can select and set a detailed detection option for skin color when a sub-category detection option for race is selected, and when a sub-category detection option for clothes is selected, the type of clothing, You can select and set the detailed detection option for clothing color, and when a sub-category detection option for belongings is selected, you can select and set detailed detection options for the type of belongings and the color of belongings, respectively.

For example, you can select items for various racial categories such as Westerners, Asians, Middle Easterners, Caucasians, Yellows, Blacks, Africans, Asians, Europeans, Americans, and South Americans, and select detailed detection options for skin color can be set by In this case, the first detection option setting unit 110 may provide a color code when a detection option related to a color is included, and select one of the provided color codes to specify the skin color of the person to be detected.

Also, the first detection option setting unit 110 may select and set a lower category detection option including at least one characteristic of an animal type, an animal size, and an animal color in the case of an upper category detection option for an animal.

For example, you can select and set detailed detection options such as wild boar, weight of about 200kg, height of 100cm, black brown, etc. If a color-related detection option is included, a color code is provided, and one of the provided color codes is selected for index finger. The skin color of the target person can be specified.

In addition, the first detection option setting unit 110, in the case of the upper category detection option for the vehicle, a vehicle model, vehicle type, vehicle size, vehicle color, and a lower category detection option including at least one characteristic of the license plate is selected. can be set.

For example, you can select and set detailed detection options such as Mojave, SUV, Large, White, **Huh****, etc., and if a color-related detection option is included, a color code is provided and one of the provided color codes is selected. By selecting one, the skin color of the person to be indexed can be specified.

In addition, in the case of the upper category detection option for the object, the first detection option setting unit 110 may receive and set a lower category detection option including at least one characteristic of an object type, object size, and object color.

For example, you can select and set detailed detection options such as bed, bench, book, chair, clock, sofa, crack, and dining table. By selecting one, the skin color of the person to be indexed can be specified.

The second detection option setting unit 120 may select and set a second detection option including a specific action or action for the control object of the first detection option, and/or an object related to the control object.

For example, through the first index finger option setting unit 110, a male in his 40s of about 180 cm in white clothes enters, roams, escapes, stops, collapses, stumbles, etc. of at least one of the actions or actions of the second index finger Options can be selected and set, and in the case of objects related to the control object, detailed detection options such as beds, benches, books, chairs, clocks, sofas, cracks, and dining tables can be selected and set as shown in FIG. 4 . Here, the object related to the control object detects the action of the set control object lying on or rising from the bed, sitting or rising on a bench, reading a book, taking off the watch, kicking or looking at it, sitting at the table or getting up from the table. It can be set as an option for

The third detection option setting unit 130 may select and set a third detection option including at least one of a detection date and time.

For example, as shown in Fig. 5, by setting a temporal range from 2017.08.01 to 2017.09.01 every month, Wednesday, and Friday, detection objects can be detected only during the corresponding date and time range.

The fourth detection option setting unit 140 may select and set the fourth detection option including the detection position. In more detail, the fourth detection option setting unit 140 may receive and set location information of at least one of a region, a region, and a range for a region or a region as the fourth detection option.

For example, as shown in FIG. 6, you can designate an area and area called Gangnam-daero and Pintel office, and you can set the range in detail by manipulating the designated area, area, and location through the map, and specific actions to be monitored , object, and detection time can also be linked and set.

Meanwhile, the CCTV detection option setting unit 100 may further include a first detection option exception setting unit 150 . The first detection option exception setting unit 150 may set the first detection option by receiving montage information about the control object through at least one of a drawing tool and a photographed image scan when the control object cannot be specified. have.

That is, the first detection option exception setting unit 150 tried to select a specific person as the detection option, but in a situation where it is difficult to specify gender, age, or facial structural characteristics, if it is difficult to set detailed detection options, sketch on paper A montage photo can be scanned and input as the first detection option, and characteristic elements that cannot be specified through the drawing tool are expressed as a picture, and the image information can be input and set as the first detection option. To this end, the first detection option exception setting unit 150 may include components such as an image scanning device, an image processing device, an image recognition device (based on deep learning technology), and a drawing tool.

In addition, the CCTV detection option setting unit 100 may further include a detection option combination setting unit 160, this detection option combination setting unit 160, the first detection option, the second detection option, the third detection When at least two groups of options and fourth detection options are simultaneously set, the first detection option and the second detection for a first condition in which the same detection options are the same and a second condition in which the same detection options are different It is possible to set a detection option combination that combines an option, the third detection option, and the fourth detection option.

As described above, the first condition means a condition in which at least two identical detection options are the same (even if the level of the upper category is the same), and the second condition is a condition in which two identical detection options are different from each other. can mean That is, when the first detection option is included in the first condition, any one of the first detection options may be a human, the other first detection option may be an animal, and the second condition includes the first detection option. In case any one of the first detection option is a person, the other first detection option may be the same as a person. However, even if both are human, the difference in sub-category or detailed detection options can be ignored.

Referring to Table 1 below, 13 detection option combinations can be created and each can be set, for example, while setting the same first detection option (A) (the first detection option is the same), each The first combination sequence set to different second to fourth detection options (B, C, D) (the second detection option is different, the third detection option is different, and the fourth detection option is different) While setting the first and second detection options (A, B) (the first detection option is the same, the second detection option is simply the same), different third and fourth detection options (C, D) (the third detection option) The second combination sequence number, .... , respectively identical to the first to third detection options (A, B, C) (the first detection option is the same, the second The twelfth combination sequence number, and the first to fourth detection options (A, A 13th combination sequence in which B, C, and D) is different from each other is possible.

<Table 1>

The CCTV selection detection unit 200 selects the CCTV images according to the detection options set through the CCTV detection option setting unit 100 and the detection option combination setting unit 160 among a plurality of CCTV images based on a deep learning algorithm. , and the corresponding CCTV images can be extracted respectively.

The CCTV selection notification unit 300 is a means for notifying the detection result of the CCTV image through the CCTV selection detection unit 200, and more specifically, the CCTV image according to the detection option set through the CCTV selection detection unit 200 is detected. Then, as shown in FIG. 7 , the detected CCTV image may be displayed in a pop-up form to inform the detection result of the CCTV image.

According to this embodiment, the detection elements according to the detection rule of 'who, when, where, what' are set in detail, and only the CCTV images according to the desired detection rule are selected using a deep learning algorithm in a pop-up form. By notifying, it is possible to reduce the number of control personnel and personnel compared to the existing ones, and to utilize the surplus time by reducing the sense of time required for control and monitoring, and to save resources and safety for the CCTV control system as it enables rapid threat detection and object detection. increase can be achieved.

What has been described above is only one embodiment for implementing a CCTV screening control system using deep learning according to the present invention, and the present invention is not limited to the above embodiment, and as claimed in the claims below Without departing from the gist of the present invention, it will be said that the technical spirit of the present invention exists to the extent that various modifications can be made by anyone with ordinary knowledge in the field to which the present invention belongs.

1000: CCTV screening control system using deep learning
100: CCTV detection option setting unit
110: first detection option setting unit
120: second detection option setting unit
130: third detection option setting unit
140: fourth detection option setting unit
150: first detection option exception setting unit
160: detection option combination setting unit
200: CCTV screening detection unit
300: CCTV selection notification unit

Claims (6)

CCTV detection option setting unit for setting a detection option including at least one of a control object to be detected in the CCTV image, a specific action / action for the control object or an object related to the control object, detection date and time, and detection location;
a CCTV selection detection unit for selectively detecting CCTV images according to a detection option set through the CCTV detection option setting unit among a plurality of CCTV images; and
A CCTV selection notification unit for notifying the detection result of the CCTV image through the CCTV selection detection unit,
The CCTV detection option setting unit,
a first detection option setting unit for selecting and setting a first detection option including a control object to be detected in the CCTV image;
a second detection option setting unit for selecting and setting a second detection option including a specific action or action for the control object of the first detection option;
a third detection option setting unit for selecting and setting a third detection option including at least one of a detection date and time; and
a fourth detection option setting unit for selecting and setting a fourth detection option including the detection position;
a first detection option exception setting unit configured to receive montage information on a control object through at least one of a drawing tool and a photographed image scan and set the first detection option when the control object cannot be specified;
When at least two groups of the first detection option, the second detection option, the third detection option, and the fourth detection option are simultaneously set, the same detection option is the same as the first condition, and the same detection option is the same. and a detection option combination setting unit for setting a detection option combination combining the first detection option, the second detection option, the third detection option, and the fourth detection option for different second conditions CCTV screening control system using deep learning.
delete According to claim 1,
The first detection option setting unit,
In the case of a higher category detection option including at least one control object among humans, animals, vehicles, and objects, and a higher category detection option for humans, at least one characteristic of gender, face, race, age, clothing, body, and belongings is detected. Select and set the included subcategory detection option,
In the case of subcategory detection options for race, select and set detailed detection options for skin color,
In the case of sub-category detection options for clothes, the detailed detection options for clothing type and clothing color are selected and set.
In the case of an upper category detection option for animals, select and set a subcategory detection option that includes at least one characteristic of animal type, animal size, and animal color,
In the case of an upper category detection option for a vehicle, select and set a lower category detection option including at least one characteristic of a vehicle model, vehicle type, vehicle size, vehicle color, and license plate,
In the case of the upper category detection option for an object, select and set a lower category detection option including at least one characteristic of object type, object size, and object color,
A CCTV screening control system using deep learning, characterized in that color-related detection options provide color codes.
According to claim 1,
The second detection option setting unit,
CCTV screening control using deep learning, characterized in that at least one of the intrusion, wandering, escaping, stopping, falling, staggering action or action of the control object, or at least one of the objects related to the control object is selected and set system.
According to claim 1,
The fourth detection option setting unit,
A CCTV screening and control system using deep learning, characterized in that the location information of at least one of the region, zone, and range for the region or zone is selected and set as the fourth detection option.
According to claim 1,
The CCTV selection notification unit,
When a CCTV image according to the detection option set through the CCTV selection detection unit is detected, the detected CCTV image is displayed in a pop-up form to inform the detection result of the CCTV image.

Images