KR102143029B1 - Artificial intelligence cctv system - Google Patents

Abstract

The present invention relates to an artificial intelligence CCTV system including: a surveillance camera unit installed in an area in which an object moves and photographing the object in the area; an AI engine unit which analyzes recognition information of the object through deep learning-based machine learning based on the image information of the object photographed by the surveillance camera unit; and a mangement server which converts and manages the database into basic information and status information for each object based on the recognition information of the object analyzed by the AI engine unit. Therefore, the present invention can efficiently and in real-time grasp the current status information of the object through an object image captured by the CCTV system, thereby performing life crime prevention.

Description

Artificial intelligence CCTV system {ARTIFICIAL INTELLIGENCE CCTV SYSTEM}

The present invention relates to an artificial intelligent CCTV system capable of performing life crime prevention by efficiently grasping current status information of an object through an object image captured by CCTV.

CCTV surveillance systems are applied in various fields for the purpose of facility management, disaster response, methods, and traffic management. As of 2013, the number of CCTV cameras installed mainly by public institutions is about 65,000 units, and is constantly increasing. With the increase of CCTV cameras, the number of videos that the controller must monitor in a limited space is increasing. In general, there are about 20*?*50 cameras suitable for monitoring by one controller. And, when the controller is in control, the monitoring ability decreases by about 45% after 12 minutes, and by about 95% after 22 minutes. Therefore, in order to operate an efficient CCTV control system, it is necessary to introduce an intelligent CCTV system based on image understanding and recognition technology.

Intelligent CCTV systems require complex video processing techniques such as object detection, tracking, and event detection. Object detection mainly requires a pedestrian or vehicle detection technique, and a flame or smoke detection technique is required depending on the situation.

The task of an intelligent CCTV system to which object detection and tracking and context recognition techniques are applied in various ways is to accurately recognize the situation by dynamic background by moving objects around the target object, time and season, and lighting changes according to changes in the weather. In order to cope with the dynamic background and changes in lighting indoors, research is being conducted to detect the situation by applying RGB-Depth cameras.

The CCTV control system can include a road security system and a daily life security system, and the road security system is installed around the boundary area of an administrative area, and uses one high-resolution camera for each lane to recognize and recognize the vehicle number of the driving vehicle. Transmit the result. The security system for living is installed around residential areas, compressing the video of a large area and transmitting it.

Registered Patent Publication No. 10-1879444 (Announcement date: 2018. 07. 17)

An object of the embodiments of the present invention for improving the above-described problem is to provide an artificial intelligent CCTV system capable of performing life crime prevention by efficiently and real-time grasping the current status information of an object through an object image captured by CCTV.

In order to achieve the above object, an artificial intelligent CCTV system according to an embodiment of the present invention includes a surveillance camera unit installed in an area where an object moves to photograph an object in the area; An AI engine unit that analyzes the recognition information of the object through deep learning-based machine learning based on the image information of the object captured by the surveillance camera unit; And a management server that converts into a database into basic information and status information for each object based on the recognition information of the object analyzed by the AI engine unit and manages it in an integrated manner.

The object includes a vehicle and a person, and the vehicle identification information includes vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information, and the identification information of the person is impression information of the person. , Clothing information, hair style information, height information, shooting count information and progress direction information may be included.

The surveillance camera unit is configured to include a plurality of CCTV cameras, and the plurality of CCTV cameras are provided with a wide-angle lens to frame-process the photographed image or video and store or transmit the image information of each captured object through a network. It can be transmitted in real time to the AI engine unit.

The AI engine unit includes a deep learning-based artificial intelligence engine, receives image information of an object transmitted by the plurality of CCTV cameras, analyzes the recognition information of the object through deep learning-based machine learning, and analyzes the Object recognition information can be transmitted to the management server through the network.

The AI engine unit extracts the recognition information of the object using a convolutional neural network layer in which a 3x3 convolution layer and a 1x1 convolution layer are combined, and operates the image information of the object as a first angle of view. In the environment, the object is detected and learned using the weight file learned with 416x416 pixels, and the object is detected and learned using the weight file learned with 320x320 pixels in the field of view operated as the second angle of view, and the first angle of view is It is characterized in that it is wider than the second angle of view.

The artificial intelligence CCTV system according to an embodiment of the present invention can perform life crime prevention by efficiently and real-time grasping the status information of the object through the object image captured by the CCTV.

According to an embodiment of the present invention, it is possible to use it in a suitable place for crime prevention and monitoring by grasping the status information of an object based on artificial intelligence.

1 is a block diagram schematically showing an artificial intelligence CCTV system according to an embodiment of the present invention.
2 is a diagram showing each configuration of an artificial intelligent CCTV system according to an embodiment of the present invention.
3 is a block diagram showing a CCTV camera in an artificial intelligent CCTV system according to an embodiment of the present invention.
4 is a block diagram showing an AI engine unit in an artificial intelligence CCTV system according to an embodiment of the present invention.

The present invention as described above will be described in detail through the accompanying drawings and embodiments.

It should be noted that the technical terms used in the present invention are only used to describe specific embodiments, and are not intended to limit the present invention. In addition, technical terms used in the present invention should be interpreted as meanings generally understood by a person having ordinary knowledge in the technical field to which the present invention belongs, unless otherwise defined in the present invention. It should not be interpreted as a meaning or an excessively reduced meaning. In addition, when the technical term used in the present invention is a wrong technical term that does not accurately represent the spirit of the present invention, it should be understood as being replaced by a technical term that can be correctly understood by those skilled in the art. In addition, the general terms used in the present invention should be interpreted as defined in the dictionary or in context before and after, and should not be interpreted as an excessively reduced meaning.

In addition, the singular expression used in the present invention includes a plural expression unless the context clearly indicates otherwise. In the present invention, terms such as “consisting of” or “comprising” should not be construed to include all of the various components, or various stages, described in the invention, including some of the components or some stages It may or may not be construed as further comprising additional components or steps.

Further, terms including ordinal numbers such as first and second used in the present invention may be used to describe elements, but the elements should not be limited by terms. The terms are only used to distinguish one component from other components. For example, the first component may be referred to as a second component without departing from the scope of the present invention, and similarly, the second component may be referred to as a first component.

Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings, but the same or similar elements are assigned the same reference numerals regardless of the reference numerals, and redundant descriptions thereof will be omitted.

In addition, in the description of the present invention, when it is determined that detailed descriptions of related known technologies may obscure the subject matter of the present invention, detailed descriptions thereof will be omitted. In addition, it should be noted that the accompanying drawings are only for easy understanding of the spirit of the present invention and should not be interpreted as limiting the spirit of the present invention by the accompanying drawings.

1 is a block diagram schematically showing an artificial intelligence CCTV system according to an embodiment of the present invention, FIG. 2 is a view showing each configuration of an artificial intelligence CCTV system according to an embodiment of the present invention, and FIG. A block diagram showing a CCTV camera in an artificial intelligence CCTV system according to an embodiment of the present invention, Figure 4 is a block diagram showing an AI engine unit in the artificial intelligence CCTV system according to an embodiment of the present invention.

1 to 2, the artificial intelligence CCTV system according to an embodiment of the present invention is a system for performing life crime prevention by efficiently and real-time grasping the status information of an object through an object image captured by CCTV. , To this end, it includes a surveillance camera unit 100, an AI engine unit 200 and a management server 300.

Meanwhile, the object in the present invention is a concept including a vehicle and a person.

In addition, the identification information of the vehicle to be described later includes vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information, and the identification information of a person includes impression information of a person, clothing information, and hairstyle. It includes information, height information, shooting count information, and progress direction information.

The surveillance camera unit 100 is a device that is installed in an area in which the object moves and photographs an object in the area, and includes a plurality of CCTV cameras 110.

The plurality of CCTV cameras 110 are provided with a wide-angle lens to frame-process the photographed image or video to be stored or transmitted, but the image information of each photographed object can be transmitted in real time to the AI engine unit 200 through a network. have.

At this time, the CCTV camera 110 may be provided in the form of a module, and the photographed image or video is processed to be stored or transmitted. For this purpose, a lens, an image sensor, an amplifier, and an analog/digital converter And an image signal processing processor.

The CCTV camera 110 may be provided with a function of recognizing various information of vehicles and pedestrians.

As shown in FIG. 3, the CCTV camera 110 used in the present artificial intelligence CCTV system includes an image capture unit 111, a lighting unit 112, a lighting control unit 113, a control unit 114, and a communication unit 115. Can include.

The image capturing unit 111 may be implemented to capture a wide area so as to identify a surrounding situation of a pedestrian as well as a driving vehicle. For example, the image capture unit 111 may have a resolution of 1.0 to 3.0 megapixels, and the image sensor may be a charge coupled device (CCD) or a complementary metal-oxide semiconductor (CMOS).

In addition, the image capturing unit 111 compresses the captured image based on the image compression standard and compresses the image data and transmits the compressed image data in real time. It may be a camera that can only handle.

At this time, if only the image data for the image captured by the image capturing unit 111 can be processed, the vehicle number recognition algorithm/face detection algorithm is applied to the image data, and the result is generated and the image for the image data Compression and real-time transmission based on compression standard technology can be performed in a software module.

Meanwhile, the software algorithm that processes data compression and real-time transmission based on the image compression standard technology is designed to be processed in a structure independent of the vehicle information recognition algorithm/face detection algorithm, and the vehicle information recognition algorithm/face detection algorithm Processing delay may not occur due to processing time. Data based on image compression standard technology The compressed image data is transmitted to the management server 300 by real-time transmission, information on the recognized vehicle is transmitted to the management server 300, and information on the recognized face is managed. It may be transmitted to the server 300.

In addition, the image capturing unit 111 may be a network camera that supports a function of compressing image data and transmitting it in real time by complying with an image extrusion standard technology for a captured image. In this case, the network camera may convert the captured image into an 8-bit gray image and then process the vehicle information recognition algorithm to generate information on the recognized vehicle. Information on the recognized vehicle is transmitted to the management server 300 and the compressed data is transmitted in real time to be linked with an operating program of the management server 300. The same method may be performed for information on the detected face.

In addition, according to an embodiment of the present invention, regardless of the type of camera constituting the image photographing unit 111, when vehicle information on a driving vehicle is recognized or a pedestrian is detected, a still image and event information are generated and stopped. Images and event information may be linked to an operating program of the management server 300. That is, according to an embodiment of the present invention, data linked to each other between the management server 300 and an external server may be generated by processing software functions of vehicle information recognition, video compression, and real-time transmission.

Meanwhile, the image capture unit 111 according to the present invention may first separate the background and the object in order to detect objects such as a vehicle and a person.

The image capturing unit 111 may recognize a portion that does not change during a critical time as a background and a portion that changes during a critical time as an object.

For example, the image capturing unit 111 may capture an image for a predetermined period of time and separate a portion that moves over a predetermined distance and a portion that does not move over a predetermined distance in the captured image. A part that does not move more than a certain distance may be recognized as a background, and a part that does not move more than a certain distance may be recognized as an object.

Alternatively, it is possible to separately set a specific frame as a background frame and input the background frame to the image capture unit 111 to perform background learning, and then separate the background and the object.

After separating the background and the object, it is determined whether or not the object appears based on the background.

The image capturing unit 111 may perform imaging by setting the widest imaging range (hereinafter, the maximum shooting range) as a basic range, and tracking the object according to the movement of the object and whether or not it is necessary to determine the object to capture the object. It can also be done.

For example, the image capturing unit 111 may include at least one auxiliary image capturing device, and at least one auxiliary image capturing device may be triggered and operated when tracking of an object is required.

When an object appears, an image of the object is captured based on tracking of the object.

Specifically, when the primary camera can capture the maximum shooting range and additional search for an object is required, the operation of at least one auxiliary image pickup device is triggered to perform a search for motion/object information of the object. I can.

The primary camera continuously performs image capturing for the maximum photographing range, and the auxiliary image capturing device may additionally perform image capturing for an object requiring precise tracking. The auxiliary image capturing device may perform imaging of an object according to the expected movement range of the object in consideration of the expected movement range of the object.

Specifically, it determines whether an existing object is a person or a vehicle, and takes an image of the overall appearance and identification information (for example, a vehicle license plate or a person's face) of the object in consideration of the object's movement path/object movement speed. Can be done.

When there are a plurality of objects to be tracked, a plurality of auxiliary cameras are allocated to each of the plurality of objects, so that each of the plurality of auxiliary cameras can perform tracking on each of the plurality of objects.

If there are many objects to be tracked, a plurality of objects may be grouped and assigned to each of a plurality of auxiliary cameras. For example, if the number of tracking target objects is 10 and there are 5 auxiliary cameras capable of tracking an object, the tracking target objects may be grouped into 5 groups and assigned to each of the 5 auxiliary cameras.

The grouping of the tracking target object may be determined based on the location of each of the plurality of tracking target objects/speed of the tracking target object. When the tracking target object is a person, grouping may be performed on the tracking target object in consideration of the current position of the person and the movement speed of the user. Grouping of the tracking target object may be performed again according to changes in the current position of the person and the movement speed of the user. For example, if object A and object B were in the same group, but object A became more adjacent to object C because the speed of object A was high, object A can switch the group to the group of object C.

The lighting unit 112 may be implemented to provide lighting according to external control in order to brighten a captured image. For example, the lighting unit 112 may have an irradiation distance of 30 m or more as an infrared light emitting diode (LED) of 730 nm to 740 nm.

The lighting control unit 113 is a component implemented to control the operation of the lighting unit 112 and may be implemented to control the intensity of lighting according to an external environment.

The control unit 114 may be implemented to recognize a vehicle number and detect a pedestrian based on an image captured by the image capturing unit 111. In addition, the control unit 114 may be implemented to control not only the image capturing unit 111 and the lighting control unit 113, but also a constituent unit constituting the CCTV camera 110. The control unit 114 may be a separate control system.

The communication unit 115 may be implemented to transmit image information captured by the image capturing unit 111 and analysis information analyzed by the control unit 114 to the management server 300.

The AI engine unit 200 is an engine installed in the surveillance camera unit 100 and/or the management server 300, and is a deep learning-based machine based on the image information of the object photographed by the surveillance camera unit 100. It is a device that analyzes the recognition information of an object through learning.

More specifically, the AI engine unit 200 is an artificial intelligence ('AI') engine based on deep learning, and receives image information transmitted from the surveillance camera unit 100, and provides deep learning. Through (Deep Learning)-based machine learning, object recognition information (for example, vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information, pedestrian impression information, clothing information, and hairstyle) Information, height information, photographing count information and progress direction information), and recognition information of the analyzed object (for example, vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information or pedestrians Impression information, clothing information, hair style information, height information, photographing count information and progress direction information) are transmitted to the management server 300 through the network.

In this case, the network provides a communication network so that communication between the surveillance camera unit 100, the AI engine unit 200 and the management server 300 can be implemented, for example, Zigbee, RF, and WiFi. ), 3G, 4G, 5G, LTE, LTE-A, wireless communication networks such as WiBro (Wireless Broadband Internet), wired communication such as PLC, JLC, Internet Web, and SNS (Social Network Service).

In other words, the AI engine unit 200 includes an artificial intelligence engine based on deep learning, receives image information of an object transmitted by a plurality of CCTV cameras 110, and provides the object through deep learning-based machine learning. The recognition information may be analyzed, and the recognition information of the analyzed object may be transmitted to the management server 300 through a network.

Hereinafter, a method of recognizing a pedestrian's face using the AI engine unit 200 will be described as an example.

Specifically, the following method is a method for determining the characteristic information of the object by extracting information on the characteristic part such as the face of the object by the CCTV, and may be performed in the management server 300 capable of communicating with CCTV or CCTV. .

The object feature information may include body information such as a pedestrian's face image and an object's height.

Information on an object (eg, a pedestrian) in the video information recorded through CCTV is continuously analyzed and may be utilized for identification of the object.

For example, the CCTV may extract primary feature information of an object by primarily analyzing image information captured for a certain period of time. Since extracting feature information of many objects consumes a lot of time and resources for processing, primary feature information can be extracted relatively simply. For example, as the primary feature information, only large feature information of a pedestrian such as gender, primary error range key information, and hair style may be extracted.

Thereafter, when a pedestrian having a similarity of the primary feature information equal to or greater than the first threshold range is photographed again a times, extraction of secondary feature information that is more detailed feature information than the primary feature information may be performed. The secondary feature information is more detailed feature information than the primary feature information, and may be information on the contour and body shape of a pedestrian's face. Through the extraction of secondary feature information, more detailed information about pedestrians whose similarity is greater than or equal to a threshold range of primary feature information can be extracted and grouped.

In the same way, when a pedestrian having a similarity of the secondary feature information equal to or greater than the second threshold range is photographed again b times, extraction of tertiary feature information, which is more detailed feature information than the secondary feature information, may be performed. The third feature information is more detailed feature information than the second feature information, and may be information about the eyes, nose, mouth, and ears of a pedestrian's face.

Through the extraction of the tertiary feature information, more detailed information on the gait in which the secondary feature information has a similarity greater than or equal to a critical range may be extracted.

Pedestrians can be classified in such a gradual manner, and characteristic information of pedestrians can be extracted. By progressively performing the analysis on the object in this manner, it is possible to effectively identify pedestrians continuously appearing while preventing unnecessary processing load.

In the case of performing such identification, separate tag information can be assigned to and managed for continuously photographed objects, and pedestrians are identified based on this tag information, and no separate analysis is performed on the object, and new Processing power can be used to analyze objects.

That is, an object that continuously passes is managed as tag information, and a detailed determination can be newly performed on an object whose primary feature information captured by CCTV is new.

If the number of new objects captured by CCTV is greater than or equal to the threshold number, the object is analyzed in a gradual manner to extract the above-described primary feature information, secondary feature information, and tertiary feature information. When the number of new objects is less than the threshold number, the object may be analyzed by extracting the second feature information or the third feature information as described above.

According to an embodiment of the present invention, the amount of processing for analysis on an object is predicted, and whether to extract primary feature information, secondary feature information or tertiary feature information for an object in consideration of the processing amount. Whether or not can be determined.

For example, when there are not many new objects relatively, the above-described threshold number of times (a, b) is adjusted relatively low, so that secondary feature information or tertiary feature information for objects that have previously extracted only primary feature information The extraction of can be performed. On the contrary, when there are relatively many new objects, the above-described threshold number (a, b) is adjusted relatively high, so that the extraction of secondary feature information or tertiary feature information for objects that previously only extracted primary feature information Deferred and preferentially, only the extraction of the primary feature information can be performed.

The above-described method may be implemented as an application or in the form of program instructions that may be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, or the like alone or in combination.

The program instructions recorded on the computer-readable recording medium are specially designed and configured for the present invention, and may be known and available to those skilled in the computer software field.

Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CDROMs and DVDs, and magnetic-optical media such as floptical disks. , And a hardware device specially configured to store and execute program instructions such as ROM, RAM, flash memory, and the like.

Examples of program instructions include not only machine language codes produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform processing according to the present invention, and vice versa.

In addition, the AI engine unit 200 extracts recognition information of the object by using a convolutional neural network layer in which a 3x3 convolution layer and a 1x1 convolution layer are combined, and the image information of the object , In the environment operated by the first angle of view, an object is detected and learned using a weight file learned with 416x416 pixels, and the object is detected and learned by using the weight file learned with 320x320 pixels at the second angle of view. I can.

In this case, it is preferable that the first angle of view is wider than the second angle of view.

More specifically, the AI engine unit 200 learns and detects data using a convolutional neural network (CNN) mainly used in the image processing field among deep learning algorithms to improve the disadvantages of image processing algorithms for detecting various objects. I can.

To this end, the AI engine unit 200 includes a detection unit 210, a recognition unit 220, and an image monitoring unit 230, as shown in FIG. 4.

The detection unit 210 may detect at least one object preset with respect to the image information captured by the surveillance camera unit 100 using a predefined detection deep learning network.

The detection unit 210 is a configuration means for detecting various objects in image information captured by the surveillance camera unit 100 based on deep learning, and uses a deep network to detect a person, detect a face, detect a vehicle, detect a license plate, etc. Can be done. Here, the detection unit 210 performs automatic object detection on the input video at regular time intervals, and can process the detected object in real time by applying a tracking technology using a deep learning network. Deep networks can be designed and trained by applying a minimization method.

In addition, the detection unit 210 automatically detects an object with respect to a surveillance camera image including an identifiable object using a detection deep network. Here, the automatically detected object (or object) may include a person, a face, a vehicle (license plate), a two-wheeled vehicle, and the like. Specifically, the detection unit 210 detects a human region using a deep learning-based network, and detects a face region using a face detection deep network within the detected human region. Likewise, the detection unit 210 may detect vehicles existing in the CCTV image by using the vehicle detection deep network, and detect a license plate area within the detected vehicle image.

In addition, the detection unit 210 automatically detects people in the input image, and provides a high-speed human detection network based on a region of interest for quickly extracting human-related information including various information such as a location, posture, and movement direction of the person. Can be used.

The detection unit 210 according to the present embodiment learns data using a convolutional neural network in order to detect robust multi-objects in various illuminance changes and deteriorated images by improving the disadvantages of the existing image processing algorithm. And can be detected.

More specifically, the detection unit 210 can process object detection with high precision by designing with 24 convolutional neural network (CNN) layers, and by combining 33 convolution layers and 11 convolution layers to reduce the amount of network computation. You can do it quickly.

In addition, the detection unit 210 may learn the image captured by the surveillance camera unit 100 in two formats of 416x416 pixels and 320320 pixels. More specifically, with respect to the input image, an object is detected and learned using a weight file learned with 416416 pixels in an environment operated with a first angle of view, and a weight file learned with 320320 pixels in a second field of view. The detection performance can be improved by detecting and learning an object using. In this case, the first angle of view may mean an environment operated with a wider angle of view than the second angle of view.

The recognition unit 220 may recognize information on a corresponding object based on a predefined recognition deep learning network and a detection result of the object detected through the detection unit 210.

In addition, the artificial intelligence image analysis of the recognition unit 220 can detect the objects of vehicles and pedestrians with respect to the input image using the learned weight file, and the type and number of objects detected on one screen are limited. none.

The image monitoring unit 230 provides the information recognized by the recognition unit 220 together with the input image so that the corresponding image can be monitored.

The artificial intelligence video analysis of the video monitoring unit 230 enables video analysis for various CCTVs that support OnVif or RTSP standard protocols, along with the function of recording CCTV camera data in a storage distribution server along with artificial intelligence video analysis. Real-time video check is possible.

In addition, when the image analysis result, vehicle number recognition, and vehicle and pedestrian detection processes are performed, the image monitoring unit 230 may display a bounding box including an object.

According to the present invention, a convolutional neural network (CNN) mainly used in the image processing field among deep learning algorithms in order to improve the shortcomings of existing image processing algorithms for detecting various objects such as vehicle and pedestrian information from images acquired from CCTV. You can learn and detect data using.

The management server 300 is a server device that converts and manages the database into basic information and status information for each object based on the recognition information of the object analyzed by the AI engine unit 200.

More specifically, the management server 300 includes recognition information of the object analyzed by the AI engine unit 200 (for example, vehicle number information, color information, model information, movement time information, movement frequency information, and progress direction information, or Pedestrian impression information, clothing information, hair style information, height information, photographing count information, and progress direction information) are received, and based on this, the status status of each object is converted into a database and managed in an integrated manner.

In this case, the management server 300 may be implemented using a PC, LM, etc., but it is natural that the management server 300 is not limited thereto and can be applied in various ways.

Hereinafter, a vehicle recognition and utilization process using the artificial intelligent CCTV system according to the present invention will be briefly described.

First, a plurality of CCTV cameras 110 detect motion of an image in a corresponding area and acquire motion information of a vehicle, a person, an object, and the like.

Next, the plurality of CCTV cameras 110 compress the acquired motion-related images to an appropriate size (for example, 1/10 to 1/15).

Next, the plurality of CCTV cameras 110 transmits the compressed motion-related images to the AI engine unit 200 in real time through wired communication (PLC, JLC, etc.).

Next, the AI engine unit 200 recognizes the object through machine learning (for example, vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information or pedestrian impression information, clothing Information, hairstyle information, height information, shooting count information and progress direction information), and recognition information of the analyzed object (for example, vehicle number information, color information, model information, moving time information, moving count information and progress) Direction information or pedestrian impression information, clothing information, hair style information, height information, photographing count information and progress direction information) are transmitted to the management server 300 through the network.

Finally, the management server 300 is the recognition information of the object analyzed by the AI engine unit 200 (for example, vehicle number information, color information, model information, movement time information, movement frequency information and progress direction information or pedestrian Impression information, clothing information, hair style information, height information, shooting count information and progress direction information) are received, and based on this, the status status of each object is converted into a database and managed in an integrated manner.

At this time, the impression information of the pedestrian may be the impression of the pedestrian (i.e., information on objects and ages such as hats, bags, glasses, etc.), the clothing information may be information on the shape and color of the clothes, and the hairstyle information is long hair, shaved hair, short hair It may be information about the shape of the head of the back.

Meanwhile, a vibration absorption prevention part of rubber material may be further installed at the lower end of the CCTV camera 110. The raw material content ratio of the vibration-absorbing part is 58% by weight of rubber, 32% by weight of carbon black, 5% by weight of antioxidant, 3% by weight of sulfur as an accelerator, and 2% by weight of resorcinol resin.

Carbon black is to increase abrasion resistance, so it is added, but if the content is less than 32% by weight, elasticity and abrasion resistance decrease, and if it exceeds 32% by weight, the content of rubber, which is the main component, is relatively small, and elasticity may decrease. , 32% by weight are mixed.

Antioxidant is 3C (N-PHENYL-N'-ISOPROPYL- P-PHENYLENEDIAMINE) or RD (POLYMERIZED 2,2,4-TRIMETHYL-1,2-DIHYDROQUINOLINE) to add 5% by weight, 5% by weight If the amount is less than 5% by weight, the product is easily oxidized, and if too much is added to exceed 5% by weight, the content of the rubber as the main component may be relatively small and the elasticity may decrease.In addition, since the price of the antioxidant is high, 5% by weight Is appropriate.

Sulfur as an accelerator is mixed in 3% by weight. If it is less than 3% by weight, 3% by weight is added since the effect of vulcanization in the heating process during molding is insignificant. If it exceeds 3% by weight, since the content of the rubber as the main component is relatively small, the elasticity may be lowered, so 3% by weight is appropriate.

Resorcinol resin is added in 2% by weight to improve adhesion.

Therefore, the present invention is reinforced with synthetic rubber having elasticity in various directions, so that the elasticity, toughness and rigidity of the CCTV camera 110 are increased, so that the durability is improved, and thus the life of the CCTV camera 110 to be mounted on various structures is increased. do.

In the above, preferred embodiments according to the present invention have been illustrated and described. However, the present invention is not limited to the above-described embodiments, and any person with ordinary knowledge in the technical field to which the present invention pertains without departing from the gist of the present invention appended to the claims will be able to implement various modifications. .

100: surveillance camera unit 110: CCTV camera
111: image photographing unit 112: lighting unit
113: lighting control unit 114: control unit
115: communication unit 200: AI engine unit
210: detection unit 220: recognition unit
230: image monitoring unit 300: management server

Claims (5)

A surveillance camera unit installed in an area in which the object moves and photographs an object in the area;
An AI engine unit that analyzes the recognition information of the object through deep learning-based machine learning based on the image information of the object captured by the surveillance camera unit; And
It includes a management server that converts into a database into basic information and status information for each object based on the recognition information of the object analyzed by the AI engine unit and manages it in an integrated manner,
The surveillance camera unit includes a plurality of CCTV cameras,
The plurality of CCTV cameras are provided with a wide-angle lens to frame-process the photographed image or video to store or transmit, but transmits the image information of each photographed object to the AI engine unit in real time through a network,
The AI engine unit includes a deep learning-based artificial intelligence engine, receives image information of an object transmitted by the plurality of CCTV cameras, analyzes the recognition information of the object through deep learning-based machine learning, and analyzes the Transmits object recognition information to the management server through the network
The CCTV camera,
A basic camera configured to set a maximum photographing range, which is the widest photographing range, as a default range to perform imaging of the maximum photographing range; And
Triggered when tracking of the object is required, and includes an image capture unit having a plurality of auxiliary cameras to perform imaging of the object,
When the number of auxiliary cameras is less than the number of objects to be tracked, a plurality of objects to be tracked are grouped and allocated to the auxiliary cameras, respectively,
The grouping of the object to be tracked is determined based on the position and speed of each object to be tracked,
An artificial intelligence CCTV system, characterized in that grouping for a corresponding object to be tracked is re-performed according to a change in the position and speed of the object to be tracked.
The method of claim 1,
The objects include vehicles and people,
Vehicle identification information includes vehicle number information, color information, model information, movement time information, movement frequency information, and direction information,
The artificial intelligence CCTV system, characterized in that the recognition information of a person includes impression information of a person, clothing information, hair style information, height information, shooting count information, and progress direction information.
delete delete According to claim 2,
The AI engine unit
The recognition information of the object is extracted using a convolutional neural network layer in which a 3x3 convolution layer and a 1x1 convolution layer are combined,
With respect to the image information of the object,
In an environment operated at the first angle of view, an object is detected and learned using a weight file learned with 416x416 pixels,
It detects and learns an object using a weight file learned with 320x320 pixels at the angle of view operated as the second angle of view,
Artificial intelligence CCTV system, characterized in that the first angle of view is wider than the second angle of view.

Images