A fire-fighting map detection and patrol method and device, computer equipment and storage medium

Abstract

The application discloses a fire-fighting map detection round-patrol method and device, computer equipment and a storage medium. The fire-fighting map detection round-patrol method is applied to a management equipment of a fire-fighting monitoring system. The round-patrol method comprises the following steps: acquiring communication addresses and communication modes of all video monitoring devices to be detected according to a fire-fighting jurisdiction; acquiring a time slice video stream in any one of the video monitoring devices according to a set time slice interval according to the communication addresses; performing map detection on the time slice video stream based on a pre-trained fire-fighting multi-target deep recognition model to obtain a monitoring result; and selecting a next video monitoring device to perform map detection in a co-prime sampling traversal mode. The application is different from a traditional simple sequential device access mode. By adopting a co-prime sampling traversal method, the sampling efficiency in a unit of time can be improved, and the scene representativeness of map detection sampling can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fire-fighting video alarm technology, and in particular to a fire-fighting image detection round-patrol method and device, computer equipment and a storage medium. BACKGROUND

[0002] In a fire safety integrated system, multiple video monitoring devices are arranged at multiple levels of fire prevention units such as buildings, parks, and streets according to fire safety requirements such as flammability, risk level, and crowd density. When the fire cloud center receives alarm signals from the front-end smoke sensor, the management personnel in the control room manually determine the fire alarm based on the real-time video monitoring pictures transmitted back from the alarm source site. However, the fire hazard warning function is not fully realized, and behaviors that violate fire safety regulations, such as occupying fire access vehicles and stacking flammable materials, cannot be timely stopped and handled.

[0003] A general solution to the above problem is an edge device based on computer vision AI processing. This type of device is placed at the camera end to perform real-time video analysis and then transmit an alarm signal back. However, this solution has drawbacks such as the need for multiple AI devices, single AI function, high cost, and high difficulty in operation and maintenance. Fire safety management departments have to bear high equipment costs and operation and maintenance costs, which is very unfavorable for the promotion of fire safety and intelligence integration.

[0004] Secondly, the determination method of the site access sequence of the existing fire-fighting equipment (such as deterministic algorithm) has a large calculation amount of the sites to be accessed in the GTP mode, resulting in low efficiency of determining the optimal site access sequence and poor results in terms of access cost (such as distance and time). SUMMARY

[0005] To address the deficiencies in the above-mentioned technologies, the present application provides a fire-fighting image detection round-patrol method, device, computer equipment, and storage medium, which replaces the sequential device access method in the prior art and uses the 'coprime sampling traversal' method to improve the sampling efficiency per unit time while ensuring the representativeness of the image detection sampling scene.

[0006] To solve the above problems, the first object of the present application is to provide a fire-fighting image detection round-patrol method applied to a management device of a fire-fighting monitoring system, which comprises:

[0007] Step S 100 : According to the fire-fighting jurisdiction, the communication address and communication method of all video monitoring devices to be image detected are obtained;

[0008] Step S 200 : According to the communication address, the time slice video stream in any one of the video monitoring devices is obtained according to the set time slice interval;

[0009] Step S300 : based on the pre-trained fire multi-target deep identification model, the time slice video stream is graphed, and a monitoring result is obtained;

[0010] Step S 400 : the next video monitoring device is selected for graphing according to the co-prime sampling traversal method.

[0011] Further, in step S 200 , the time slice video stream is a plurality of frames of images after the current time of the video monitoring device, and the time slice length is less than 10 seconds.

[0012] Further, the fire monitoring system comprises a video monitoring device, a fire alarm sensor, a fire alarm controller and a cloud center server, and the video monitoring device, the fire alarm sensor and the fire alarm controller are in communication connection with the cloud center server.

[0013] Further, the fire multi-target deep identification model comprises a fire truck passage detection model, a motor vehicle detection model, an electric vehicle detection model, a evacuation passage detection model and a flammable material detection model.

[0014] Further, in step S 400 , the co-prime sampling traversal method comprises:

[0015] S 410 : the total number N of all video monitoring devices is counted, and the video monitoring devices are numbered in ascending order V i (i=0~N-1);

[0016] S 420 : the sampling interval set X={X j} is calculated, wherein X j is co-prime with N, and X j <N;

[0017] S 430 : X j is randomly selected from the sampling interval set X, and a co-prime sampling traversal is performed once.

[0018] S 440 : after one sampling traversal, X j is randomly selected again, and the next sampling traversal is performed.

[0019] Further, in step S 430 , the co-prime sampling traversal once comprises:

[0020] S 431 : the current sampling interval is determined as X j ;

[0021] S432 : set the starting image detection task number k as a value p, execute image detection of the video monitoring device V k ;

[0022] S 433 : update k=(k+X j ) % N, execute image detection of the video monitoring device V k , until termination when k=p again, and complete image detection of all video monitoring devices;

[0023] Wherein %, remainder operation, 0<=k<=N-1, and k is an integer.

[0024] A second object of the present application is to provide a fire-fighting image detection device, which adopts the fire-fighting image detection round-patrol method described above, and the device comprises:

[0025] A communication module, configured to receive time-slice video streams sent by a plurality of video monitoring devices;

[0026] An image detection module, configured to determine a monitoring result according to whether the time-slice video streams satisfy a fire-fighting multi-target recognition algorithm model;

[0027] A traversal module, configured to determine an image detection traversal order of the plurality of video monitoring devices according to an order of the video monitoring devices in a fire-fighting management system.

[0028] A third object of the present application is to provide a computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the fire-fighting image detection round-patrol method described above.

[0029] A fourth object of the present application is to provide a computer-readable storage medium, which stores a computer program, wherein the computer program is executable on a processor to implement the fire-fighting image detection round-patrol method described above.

[0030] Compared with the prior art, the present application has remarkable advantages and beneficial effects, which are embodied in the following aspects:

[0031] 1. The fire-fighting map detection round patrol method in the present application, first, according to the fire-fighting jurisdiction, the communication address and the communication mode in all video monitoring devices to be detected are obtained; based on the obtained communication address, the time slice interval is set, the time slice video stream in any one video monitoring device in the fire-fighting jurisdiction can be obtained; then through the pre-trained fire-fighting multi-target deep recognition model, the time slice video stream obtained can be detected, and the monitoring result of the video monitoring device under the communication address is obtained; the next video monitoring device is selected for map detection according to the co-prime sampling traversal method; finally, through the "co-prime sampling traversal" method, more representative fire-fighting monitoring point position pictures can be collected in one round of patrol, and the sampling operation can be automatically adapted according to the total number of monitoring points, which greatly avoids the time slice waiting hidden danger caused by the fixed mode of round patrol sampling sequence, and effectively reduces the working strength of the fire control room management personnel, reduces the overall cost of the system and the operation and maintenance cost of the front-end equipment.

[0032] 2. Compared with the prior art, the specific map detection function can also be customized and deployed at any time, and compared with the scheme of deploying a single algorithm in the front-end camera (edge device), the present application has significant technical advantages and cost advantages. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 is a flowchart of the fire-fighting map detection round patrol method in the embodiment of the present application;

[0034] Figure 2 is a structural schematic diagram of the fire-fighting map detection device in the embodiment of the present application;

[0035] Figure 3 is a structural schematic diagram of the computer device in the embodiment of the present application;

[0036] Figure 4 is a flowchart of the co-prime sampling traversal method in the embodiment of the present application;

[0037] Figure 5 is an application scenario schematic diagram of the fire-fighting map detection round patrol method in the embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0039] It should be noted that, in this article, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0040] Please refer to Figures 1-5 The embodiment of the application provides a fire-fighting map detection and patrol method, which is applied to a management device of a fire-fighting monitoring system, and the map detection and patrol method comprises the following steps.

[0041] Step S 100 According to the fire-fighting jurisdiction, the communication address and the communication mode of all video monitoring devices to be detected are obtained.

[0042] It can be understood that, with the rapid development of the video monitoring industry, more and more video channels (each camera corresponds to a video channel) are connected to the video management system for management. In the process of managing the videos of various video channels in the video management system, how to ensure the integrity and availability of the videos of various video channels becomes crucial.

[0043] Therefore, the fire-fighting cloud center in the embodiment can obtain the live video stream of a certain path by connecting the video monitoring devices associated with the front-end fire-fighting monitoring points through the network.

[0044] Step S 200 According to the communication address, the time slice video stream in any one video monitoring device is obtained according to the set time slice interval.

[0045] It should be noted that the video monitoring device can send the video segments to the target multicast address according to the order of the video segments corresponding to the target video stream based on the time distribution information of each video segment in the target video stream information, so as to ensure the continuity and integrity of the video segments read by the client from the target multicast address.

[0046] Step S 300 Based on the pre-trained fire-fighting multi-target deep recognition model, the obtained time slice video stream is detected to obtain a monitoring result.

[0047] The graph detection algorithm is deployed in a cloud center server, and calculation and analysis are performed by an algorithm in the cloud center server; when the calculation capability of the cloud center server is limited, the returned video stream is calculated and analyzed in turn according to time slices.

[0048] Step S 400 The next video monitoring device is selected according to the mutually prime sampling traversal mode.

[0049] It can be understood that the video monitoring devices with adjacent numbers in the integrated safety and security system generally have geographical spatial correlation. For example, the monitoring fields of view of multiple cameras in the same warehouse will have a certain degree of overlap or continuity. If the default round-robin mode is used, the live scene of the point with a late number will be artificially delayed, which causes the disadvantage of a time slice blind area of fire safety hazards in some areas.

[0050] Compared with the sequential device access mode in the prior art, the embodiment of the present application can improve the sampling efficiency per unit time while ensuring the scene representativeness of graph detection sampling by using the'mutually prime sampling traversal' method.

[0051] Specifically, in the embodiment of the present application, the fire monitoring system includes video monitoring devices, fire alarm sensors, fire alarm controllers and a cloud center server, and the video monitoring devices, the fire alarm sensors and the fire alarm controllers are all in communication connection with the cloud center server.

[0052] Relying on a public network or a private network, the monitoring center of the system and the fire department are networked, the fire alarm information, the operation state information of building fire facilities and the fire safety management information of the networked users are received, processed and managed, the confirmed fire alarm information is sent to the fire communication command center or other alarm receiving centers, the fire department is provided with query, and information services are provided for the networked users.

[0053] Specifically, in step S 200 In the embodiment, the time slice video stream is a plurality of frames of images after the current time of the video monitoring device, and the time slice length is less than 10 seconds.

[0054] The video stream processing device can cut a plurality of video streams into a plurality of video stream segments, send each video stream segment and its metadata information to a corresponding processing module for distributed processing to obtain a plurality of frames of images, and then sort the plurality of frames of images. Such a mode improves the video stream processing efficiency, reduces the phenomenon of lag and delay in the video stream processing process, can obtain the processing result of the video stream in time or even in real time, and can be used in scenes with high real-time requirements such as security, live broadcast and face recognition.

[0055] Considering the shooting delay of the video monitoring device, generally, several frames of images are selected after the starting moment of the video monitoring device, and the time slice length is not too long, otherwise, the calculation amount is increased, and the site access efficiency is affected.

[0056] Specifically, in the embodiment of the present application, the fire-fighting multi-target depth recognition model includes a fire truck passage detection model, a motor vehicle detection model, an electric vehicle detection model, an evacuation passage detection model and a combustible material detection model.

[0057] The fire truck passage detection model in the embodiment detects obstacles in the fire passage through sensors arranged by infrared, ultrasonic wave and the like, and sends an alarm when an obstacle is detected.

[0058] The motor vehicle detection model generally simulates different vehicle types by changing the vehicle model, and can detect the actual size and weight of the vehicle according to a certain scale reduction, so as to use a more realistic vehicle model for motor vehicle detection line simulation, thereby improving the accuracy and reliability of simulation.

[0059] The evacuation passage detection model generally generates a plane evacuation route of the evacuation space according to the contour data of the space, and maps the plane evacuation route to the ground of the three-dimensional evacuation space when detecting a high-difference passage in the evacuation space, so as to automatically generate a more accurate three-dimensional space evacuation route and improve the drawing efficiency of the evacuation route.

[0060] The combustible material detection model generally detects the concentration of flammable gas in the air by using a sensor, converts the detection result into an electrical signal, and then analyzes and processes the signal through a processor.

[0061] Please refer to Figure 4 in the drawing, in step S 400 Among them, the co-prime sampling traversal method includes:

[0062] S 410 : count the total number N of all video monitoring devices, and incrementally number the video monitoring devices V i (i=0~N-1);

[0063] S 420 : calculate the sampling interval set X={X j}, wherein X j is co-prime with N, and X j <N;

[0064] S 430 : randomly select X j from the sampling interval set X, and perform a co-prime sampling traversal;

[0065] S 440: once sampling traversal is completed, randomly select X j , and then perform next sampling traversal.

[0066] In the correlation detection application of signals, with the increase of the frequency and bandwidth of the detected signals, the sampling frequency is required to be greatly increased, and the amount of data to be processed is also increased, which causes the difficulty in actual system implementation. In the embodiment of the present application, the detected signals are branched by using the coprime sampling traversal method, and coprime sampling is adopted, so that the required sampling frequency is greatly reduced on the basis of ensuring the detection performance under low signal-to-noise ratio. Theoretical analysis and experimental simulation show that the coprime sampling traversal method not only has good detection performance, but also greatly reduces the required sampling rate and the complexity of the algorithm.

[0067] Specifically, in the embodiment of the present application, the once coprime sampling traversal includes:

[0068] S 431 : determine that the current sampling interval is X j ;

[0069] S 432 : set the starting image detection task number k as a value p, and perform image detection of the video monitoring device V k ;

[0070] S 433 : update k=(k+X j ) % N, and perform image detection of the video monitoring device V k , until k=p again, and the image detection of all video monitoring devices is completed.

[0071] Wherein, % is the remainder operation, 0≤k≤N-1, and k is an integer.

[0072] As shown in Figure 4 , in a possible embodiment, there are 13 video monitoring devices in the monitoring area, numbered 0-12, the starting number p is set as 0, the sampling interval set is {2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12}, a sampling interval 2 is randomly selected, the order of the access value k according to the coprime sampling traversal method is 0, 2, 4, 6, 8, 10, 12, 1, 3, 5, 7, 9, 11 (after the interval 2, the number k returns to 0=p), and the image detection of all video monitoring devices is completed once.

[0073] Then, a sampling interval 4 is randomly selected from the sampling interval set, and the next traversal is performed, the order of the access value k according to the coprime sampling traversal method is 0, 4, 8, 12, 3, 7, 11, 2, 6, 10, 1, 5, 9, and the second traversal is completed, and the subsequent is carried out in this way.

[0074] Please see Figure 2 As shown, embodiments of the present invention also provide a fire detection image device, employing the fire detection image patrol method described above. The fire detection image device includes:

[0075] Communication module 201 is used to receive time-slice video streams sent by multiple video surveillance devices;

[0076] The image detection module 202 is used to determine the monitoring results based on whether the time-slice video stream meets the fire multi-target recognition algorithm model;

[0077] Traversal module 203 is used to determine the image traversal order of multiple video surveillance devices based on their order in the fire management system.

[0078] Combination Figure 3 As shown, this embodiment of the invention also provides a computer device that can perform the function of managing equipment in a fire protection system. The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The memory 302 is used to store program instructions; the processor 301 is used to call the program instructions stored in the memory 302. When the processor executes the computer program, it performs the following steps:

[0079] Acquire time-slice video streams sent by multiple video surveillance devices;

[0080] Based on the acquired video stream, determine whether it meets the fire multi-target recognition algorithm model and determine the monitoring results;

[0081] The order of image traversal for multiple video surveillance devices is determined based on their order in the fire management system.

[0082] Please refer to Figure 5 The diagram illustrates an application scenario of an embodiment of the present invention. In one possible embodiment, multiple video surveillance devices 102-105 are communicatively connected to a cloud central server 101. The video surveillance devices collect video data within the monitored area and send the video data to the cloud central server 101. The cloud central server 101 obtains the communication addresses and communication methods of all video imaging devices to be monitored through a communication module; then, according to a set time-slice interval, it acquires the time-slice video stream of a specific video imaging device.

[0083] Specifically, embodiments of the present invention also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the fire-fighting image reconnaissance patrol method as described above.

[0084] The computer-readable storage medium of this invention uses time-slice video streams sent by multiple video surveillance devices as reference conditions. Through an image detection module, it determines whether the time-slice video streams meet the fire multi-target recognition algorithm model and determines the monitoring results. Based on the order of the video surveillance devices in the fire management system, it determines the image detection traversal order of multiple video surveillance devices and adopts the 'coprime sampling traversal' method. This can improve the sampling efficiency per unit time while ensuring the scene representativeness of the image detection sampling. In cases with a large number of video surveillance points, it avoids the drawback of human delay in the analysis of the on-site images of points with later numbering in the traditional default round-robin method, which would cause time-slice blind spots for fire hazards in certain areas.

[0085] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM, electrically programmable ROM, electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM, dynamic RAM, synchronous DRAM, dual data rate SDRAM, enhanced SDRAM, synchronous link DRAM, memory bus direct RAM, direct memory bus dynamic RAM, and memory bus dynamic RAM, etc.

[0086] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A fire monitoring image patrol method, applied to the management equipment of a fire monitoring system, characterized in that, The image detection and polling method includes: Step S 100 Based on the fire jurisdiction, obtain the communication addresses and communication methods of all video surveillance devices to be investigated. Step S 200 Based on the communication address and according to the set time-slice interval, obtain the time-slice video stream from any of the video surveillance devices; Step S 300 Based on a pre-trained fire-fighting multi-target deep recognition model, image analysis is performed on the acquired time-slice video stream to obtain monitoring results; Step S 400 : Select the next video surveillance device for image detection using the coprime sampling traversal method; The coprime sampling traversal method includes: S 410 : Count the total number N of all video surveillance devices, and assign incremental numbers V to each video surveillance device. i (i=0~N-1); S 420 : Calculate the sampling interval set X={X j }, where X j X is coprime to N and X is coprime to N. j <N; S 430 : Randomly select X from the sampling interval set X j And perform a coprime sampling traversal; The first coprime sampling traversal includes: S 431 : Determine the current sampling interval as X j ; S 432 Set the initial image detection task number k to the value p, and execute image detection on video surveillance device Vk; S 433 Update k=(k+X) j )%N, execute image detection of video surveillance device Vk until k=p again, and then terminate, completing one image detection of all video surveillance devices; Where: % is the modulo operation, 0≤k≤N-1, and k is an integer; S 440 After one sampling traversal is completed, X is randomly selected again. j Then proceed with the next sampling iteration.

2. The fire protection image reconnaissance patrol method according to claim 1, characterized in that, In step S 200 The time-slice video stream consists of several frames of images following the current moment of the video surveillance device, and the time slice length is less than 10 seconds.

3. The fire protection image reconnaissance patrol method according to claim 1, characterized in that, The fire monitoring system includes video surveillance equipment, fire alarm sensors, fire alarm controllers, and a cloud center server, and the video surveillance equipment, fire alarm sensors, and fire alarm controllers are all communicatively connected to the cloud center server.

4. The fire protection image reconnaissance patrol method according to claim 1, characterized in that, The fire-fighting multi-target deep recognition model includes a fire truck access detection model, a motor vehicle detection model, an electric vehicle detection model, an evacuation route detection model, and a flammable material detection model.

5. A fire-fighting image reconnaissance device, employing the fire-fighting image reconnaissance patrol method according to any one of claims 1-4, characterized in that, The device includes: The communication module is used to receive time-slice video streams sent by multiple video surveillance devices; The image detection module is used to determine the monitoring results based on whether the time-slice video stream meets the fire multi-target recognition algorithm model; The traversal module is used to determine the image traversal order of the multiple video surveillance devices according to their order in the fire management system.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the fire-fighting image reconnaissance patrol method as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fire reconnaissance patrol method according to any one of claims 1 to 4.

Images