CN114186735A - Fire-fighting emergency illuminating lamp layout optimization method based on artificial intelligence - Google Patents
Fire-fighting emergency illuminating lamp layout optimization method based on artificial intelligence Download PDFInfo
- Publication number
- CN114186735A CN114186735A CN202111507302.4A CN202111507302A CN114186735A CN 114186735 A CN114186735 A CN 114186735A CN 202111507302 A CN202111507302 A CN 202111507302A CN 114186735 A CN114186735 A CN 114186735A
- Authority
- CN
- China
- Prior art keywords
- fire
- emergency lighting
- fighting
- sub
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 16
- 238000005286 illumination Methods 0.000 claims description 53
- 238000012549 training Methods 0.000 claims description 28
- 238000012545 processing Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 6
- 231100001261 hazardous Toxicity 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 5
- 238000010586 diagram Methods 0.000 claims description 2
- 230000008014 freezing Effects 0.000 claims description 2
- 238000007710 freezing Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- ZRHANBBTXQZFSP-UHFFFAOYSA-M potassium;4-amino-3,5,6-trichloropyridine-2-carboxylate Chemical compound [K+].NC1=C(Cl)C(Cl)=NC(C([O-])=O)=C1Cl ZRHANBBTXQZFSP-UHFFFAOYSA-M 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Development Economics (AREA)
- Computing Systems (AREA)
- Educational Administration (AREA)
- Computer Security & Cryptography (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Circuit Arrangement For Electric Light Sources In General (AREA)
Abstract
The invention relates to the technical field of fire fighting equipment deployment, in particular to a fire fighting emergency illuminating lamp layout optimization method based on artificial intelligence. The method comprises the following steps: acquiring a building information model and a global image of a target building space; obtaining a fire emergency lighting probability map corresponding to the global image according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability of each pixel point of the global image; inputting the fire emergency lighting probability map and the number of fire emergency lighting lamps into a trained fire emergency lighting lamp layout optimization network to generate a fire emergency lighting lamp layout scheme; and adjusting the original layout of the corresponding fire-fighting emergency lighting lamps in the building information model according to the fire-fighting emergency lighting lamp layout scheme. The invention optimizes the layout of the fire-fighting emergency lighting lamps in the target building space by utilizing the fire-fighting emergency lighting lamp layout optimization network so as to ensure that the layout of the fire-fighting emergency lighting lamps can meet the actual requirement.
Description
Technical Field
The invention relates to the technical field of fire fighting equipment deployment, in particular to a fire fighting emergency illuminating lamp layout optimization method based on artificial intelligence.
Background
The fire-fighting emergency lamp is an important lighting device in fire safety management, provides lighting for personnel evacuation and fire-fighting operation, and is arranged for guiding trapped personnel to evacuate or develop fire-fighting rescue actions after a normal lighting power supply is cut off when a fire disaster occurs, so that the layout of the fire-fighting emergency lamp plays a vital role in guaranteeing the life safety of personnel.
In actual life, along with the time lapse, the personnel distribute and the layout characteristics of different regions in the building can constantly change, for example there is the position of potential safety hazard in the room, the position of fire-fighting equipment all can change, consequently when the conflagration breaing out, the region that needs the illumination also needs corresponding adjustment, and the fire emergency light overall arrangement often is fixed among the actual process, consequently can't effectually satisfy actual demand.
Disclosure of Invention
In order to solve the problem that the layout of the fire-fighting emergency lamps cannot effectively meet the actual requirements, the invention aims to provide a method for optimizing the layout of the fire-fighting emergency lamps based on artificial intelligence, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a method for optimizing a layout of fire emergency lighting lamps based on artificial intelligence, including the following steps:
acquiring a building information model of a target building space and a global image of the target building space;
obtaining a corresponding fire emergency lighting probability map according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to each pixel point in the global image;
inputting the fire emergency lighting probability map and the number of fire emergency lighting lamps into a trained fire emergency lighting lamp layout optimization network to generate a corresponding fire emergency lighting lamp layout scheme;
and adjusting the original layout of the corresponding fire-fighting emergency lighting lamps in the building information model according to the fire-fighting emergency lighting lamp layout scheme.
Preferably, the fire emergency lighting lamp layout optimization network includes a generator sub-network and a discriminator sub-network, and the process of training the fire emergency lighting lamp layout optimization network includes:
acquiring a lighting area training set, wherein the lighting area training set comprises actual lighting areas of fire-fighting emergency lighting lamps at different positions and randomly generated lighting areas; setting corresponding first label data for each lighting area in a lighting area training set, wherein the first label data comprises: actual and non-actual compliance;
recording the lighting area in which the first label data accords with the reality as a first lighting area, and taking the position information of the fire-fighting emergency lighting lamp corresponding to the first lighting area as second label data of the first lighting area; recording the illumination area in which the first label data does not conform to the reality as a second illumination area, and using the set position information as second label data of the second illumination area;
training a discriminator sub-network according to a lighting area training set, first label data and second label data, wherein the first label data are label data of a first classifier of the discriminator sub-network, the second label data are label data of a second classifier of the discriminator sub-network, and the discriminator sub-network is used for obtaining position information of a fire-fighting emergency lighting lamp corresponding to a lighting area;
freezing the network parameters of the trained discriminator sub-network, and training a generator sub-network by using the trained discriminator sub-network and a historical fire emergency lighting probability map, wherein the generator sub-network is used for generating a lighting area according with the actual situation.
Preferably, the loss function of the generator subnetwork is calculated as:
therein, Loss2For generating loss function values of a sub-network of devices, ROI11 st illumination area, ROI, generated for the generator subnetwork22 nd illumination region, ROI, generated for a generator subnetworkNNth illumination area, ROI, generated for a generator subnetworkfTo generate the f-th illumination area for the generator sub-network,
as the abscissa of the center point of the f-th illumination area,
is the ordinate of the center point of the f-th illumination area,
is the lighting probability corresponding to the center point of the f-th lighting area, w (x, y) is the lighting probability of the pixel point with x as the abscissa and y as the ordinate in the f-th lighting area,
the result of the discrimination corresponding to the f-th illumination region output by the first classifier of the discriminator sub-network is N, the number of illumination regions generated by the generator sub-network, and e is a natural constant.
Preferably, the process of generating the corresponding layout scheme of the fire-fighting emergency lighting lamps comprises the following steps of inputting the fire-fighting emergency lighting probability map and the number of the fire-fighting emergency lighting lamps into a trained fire-fighting emergency lighting lamp layout optimization network:
dividing the global image into a plurality of sub-regions according to the positions of all rooms in the building information model;
dividing the fire emergency lighting probability map according to each corresponding subregion in the global image to obtain a sub fire emergency lighting probability map corresponding to each subregion;
and inputting the sub-fire emergency lighting probability map corresponding to each subregion in the global image and the number of the fire emergency lighting lamps corresponding to each subregion into a trained fire emergency lighting lamp layout optimization network to generate a fire emergency lighting lamp layout scheme corresponding to each subregion.
Preferably, according to fire emergency lighting lamp overall arrangement scheme adjusts the fire emergency lighting lamp overall arrangement that corresponds in the building information model, include:
acquiring original position information of each fire-fighting emergency lighting lamp in each subregion according to the layout information of the fire-fighting emergency lighting lamps in the building information model;
constructing a corresponding bipartite graph according to the original position information of each fire-fighting emergency lighting lamp in each subarea and the target position information of each fire-fighting emergency lighting lamp corresponding to the fire-fighting emergency lighting lamp layout scheme corresponding to each subarea;
processing bipartite graphs corresponding to the sub-regions by using a KM algorithm to obtain a matching pair set of original position information and target position information of the fire-fighting emergency lighting lamps in the sub-regions;
and calculating the reasonable degree of the original layout of the fire-fighting emergency illuminating lamps corresponding to the sub-regions according to the matching pair set corresponding to the sub-regions, judging whether the reasonable degree of the original layout of the fire-fighting emergency illuminating lamps corresponding to the sub-regions is smaller than a reasonable degree threshold value, and if so, adjusting the original positions of the fire-fighting emergency illuminating lamps in the corresponding sub-regions to the corresponding target positions.
Preferentially, according to the corresponding personnel distribution probability, fire safety hidden danger degree probability, fire fighting equipment distribution probability of each pixel point of the global image, obtain corresponding fire emergency lighting probability map, include:
acquiring continuous multi-frame global images, and recording as a global image sequence;
carrying out binarization processing on each image according to the position of a personnel key point corresponding to each image in the global image sequence to obtain a personnel distribution diagram sequence; calculating the average value of the pixel values of all pixel points corresponding to all positions in the personnel distribution map sequence, and recording the average value as the personnel distribution probability corresponding to the pixel points of all positions;
inputting the global image sequence into a semantic segmentation network to obtain a fire fighting equipment mask image and a dangerous goods mask image;
carrying out binary processing on each image in the global image sequence according to the fire fighting equipment shade image to obtain a fire fighting equipment shade binary image sequence; calculating the average value of pixel values of all pixel points corresponding to all positions in the fire fighting equipment shade binary image sequence, and recording the average value as the fire fighting equipment distribution probability corresponding to the pixel points of all the positions;
carrying out binary processing on each image in the global image sequence according to the dangerous article mask image to obtain a dangerous article mask binary image sequence; calculating the average value of the pixel values of all pixel points corresponding to all positions in the binary image sequence of the hazardous article shade, and recording the average value as the distribution probability of the fire safety hidden danger corresponding to the pixel points of all the positions;
obtaining the lighting probability corresponding to each position pixel point according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire fighting equipment distribution probability corresponding to each position pixel point;
and taking the illumination probability corresponding to each position pixel point as the pixel value of the corresponding position pixel point to obtain a fire-fighting emergency illumination probability map.
Preferably, the calculation formula for calculating the reasonable degree of the layout of the original fire-fighting emergency lighting lamps corresponding to each subregion is as follows:
wherein, Score is the reasonable degree of original layout of fire emergency lighting lamps in the sub-region, L2(j) is the L2 distance between the original position information and the target position information in the jth matching pair, and M is the total number of matching pairs in the matching pair set corresponding to the sub-region.
The embodiment of the invention has the following beneficial effects:
according to the method, the obtained building information model in the target building space and the global image of the target building space are analyzed, then the corresponding fire emergency lighting probability map is obtained according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to the pixel points at all positions, and finally the fire emergency lighting probability map and the number of the fire emergency lighting lamps are input into a trained fire emergency lighting lamp layout optimization network, so that a fire emergency lighting lamp layout scheme for adjusting the original layout of the fire emergency lighting lamps is generated. The invention optimizes the layout of the fire-fighting emergency lighting lamps in the target building space by utilizing the fire-fighting emergency lighting lamp layout optimization network so as to ensure that the layout of the fire-fighting emergency lighting lamps can meet the actual requirement.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for optimizing a layout of a fire emergency lighting lamp based on artificial intelligence according to an embodiment of the present invention.
Detailed Description
In order to further illustrate the technical means and functional effects of the present invention adopted to achieve the predetermined invention purpose, the following detailed description will be made on a fire emergency lighting lamp layout optimization method based on artificial intelligence according to the present invention with reference to the accompanying drawings and preferred embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the fire-fighting emergency illuminating lamp layout optimization method based on artificial intelligence in detail with reference to the accompanying drawings.
An embodiment of a fire-fighting emergency lighting lamp layout optimization method based on artificial intelligence comprises the following steps:
as shown in fig. 1, the method for optimizing the layout of the fire emergency lighting lamp based on artificial intelligence in this embodiment includes the following steps:
step S1, a building information model of the target building space and a global image of the target building space are obtained.
In order to obtain a fire emergency lighting probability map corresponding to each region in a current target building space, the present embodiment obtains a building information model of the current target building space and a global image of the target building space, where the target building space is a building space analyzed by the present embodiment, and specifically includes:
in the embodiment, a Building Information Model (BIM) of a target building space is firstly obtained, and then RGB images in each camera view range are obtained by using monitoring cameras arranged in the target building space, wherein the position of each camera is fixed, and the pose of each camera is an oblique top view angle; in the embodiment, in consideration of the fact that the visual field ranges of the two cameras have an overlapping area, the embodiment performs image stitching on the acquired RGB images to obtain a global image corresponding to the target building space. According to the embodiment, the one-to-one correspondence between the position information in the global image and the position information in the building space can be realized by means of the building information model, and meanwhile, the layout information of the current fire-fighting emergency lamp in the target building space can also be obtained.
And step S2, obtaining a corresponding fire emergency lighting probability map according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to each pixel point in the global image.
The embodiment considers that the internal layout in the target building space can change along with the change of time, so the embodiment performs rationality analysis on the layout of the fire emergency lighting lamp in the current target building space at fixed time intervals, the preferred fixed time of the embodiment is one month, and the specific fixed time can be set according to actual needs.
In order to optimize the original layout of the fire-fighting emergency lighting lamp in the target building space, the embodiment analyzes the acquired global image to obtain a fire-fighting emergency lighting probability map corresponding to the global image. The fire emergency lighting probability map reflects the probability that each location in the target building space needs to be illuminated. The process of specifically obtaining the fire emergency lighting probability map corresponding to the global image is as follows:
in the embodiment, it is considered that due to the flow of people, the fire fighting equipment and dangerous goods in the acquired images can be shielded, and the distribution of people at different positions in the building space is different at different times, so that the embodiment acquires continuous multi-frame global images, records the continuous multi-frame global images as a global image sequence, and analyzes the image sequence to obtain the fire emergency lighting probability corresponding to each position pixel point. This implementation considers because normal lighting power is cut off when the conflagration takes place, the position of people's unable accurate discernment hazardous articles and fire-fighting equipment, consequently need fire control emergency lighting lamp to the specific area light up arousing people's concern, in addition in carrying out when sparse because people can't see everyone's position clearly, can lead to trampling dangerous emergence such as incident very easily, consequently this embodiment is through carrying out the analysis to personnel's distribution situation, fire-fighting equipment distribution situation and hazardous articles distribution situation three aspect, obtain the fire control emergency lighting probability picture that global image corresponds, specifically be:
first, in this embodiment, each image in the global image sequence is input into the keypoint detection network, and gaussian hotspots corresponding to the keypoints of both feet of the human body in the global image are output. The structure of the key point detection network in the implementation is Encoder-Decoder, the Encoder performs feature extraction on an input image to obtain a feature map, and then the obtained feature map is input to the Decoder to perform upsampling to obtain a two-foot key point Gaussian hot spot image which is as large as the input image. The key point detection network in this embodiment may adopt existing CPM and CPN networks.
And after Gaussian hot spots of key points of the human body of each image in the global image sequence are obtained, processing each image in the global image sequence to obtain a personnel distribution map corresponding to each image, wherein the personnel distribution map is the distribution condition of the personnel in the image. Specifically, in this embodiment, each gaussian hot spot in the global image output by the key point detection network is processed by using the Soft-argmax function, so as to obtain the position information of each individual key point. Then, according to the position information of each human body key point in the global image, the embodiment performs binarization processing on the global image, that is, the pixel value of the pixel point at each human body key point in the global image sequence is set to 1, and the pixel points in other regions are set to 0, so as to obtain a personnel distribution map corresponding to the global image. The size of the sequence of the personnel distribution maps obtained in this embodiment is [ W, H, T ], where W, H are the width and length of each personnel distribution map, and T is the sequence length, i.e., the number of frames of the global image acquired in a fixed time period.
In this embodiment, each position in the personnel distribution map sequence corresponds to a personnel distribution binary sequence of 1 row and T columns, that is, a binary sequence formed by pixel values of each pixel point corresponding to the corresponding position in the personnel distribution image sequence reflects the personnel distribution condition at the position. In this embodiment, an average value is calculated for each element in the binary sequence corresponding to each position, the obtained value is the personnel distribution probability of the pixel point at the corresponding position, and the personnel distribution probability of the pixel point at the position (x, y) is recorded as w1(x, y), the greater the probability of personnel distribution indicates that the probability of personnel at the corresponding position is greater, and the greater the probability of personnel being present at the corresponding position is, the greater the probability of personnel being illuminated by the fire emergency lighting lamp.
Secondly, inputting each image in the global image sequence into a semantic segmentation network to obtain a fire fighting equipment Mask image Mask corresponding to each image1Mask image Mask for dangerous goods2；Mask1Responding to the distribution of fire-fighting equipment, Mask2The distribution of dangerous goods is reacted. The semantic segmentation network in the implementation is an Encoder-Decoder structure, and the existing semantic segmentation networks such as UNet and Deeplab V3 can be selected in the actual implementation process. In this embodiment, the fire fighting equipment includes equipment such as fire extinguisher, indoor fire hydrant, the hazardous articles are objects that take place the fire easily such as block terminal, computer.
The implementation carries out binarization processing on the fire fighting equipment shade image corresponding to each image in the global image sequence, sets the pixel value of the pixel point of the fire fighting equipment area in each image to be 1, and sets the pixel values of other areas to be 0 to obtain the fire fighting equipment shade image sequence; the implementation carries out binarization processing on the dangerous goods mask image corresponding to each image in the global image sequence, sets the pixel value of the pixel point of the dangerous goods area in each image to be 1, and sets the pixel values of other areas to be 0, so as to obtain the dangerous goods mask image sequence.
Each position in the fire fighting equipment shade image sequence corresponds to a fire fighting equipment binary sequence with 1 row and T columns, the average value of each element in the fire fighting equipment binary sequence corresponding to each position in the fire fighting equipment shade image sequence is calculated to obtain the fire fighting equipment distribution probability corresponding to each position pixel point, and the fire fighting equipment distribution probability of the pixel point at the position (x, y) is recorded as w2(x, y); the larger the distribution probability value of the fire fighting equipment is, the larger the probability that the fire fighting equipment exists in the corresponding position is, and the larger the probability that fire fighting operation occurs is, so that the corresponding position needs to be illuminated by a fire fighting emergency illuminating lamp.
In this embodiment, according to the method for obtaining the distribution probability of the fire fighting equipment corresponding to each position pixel point, the hazardous article shade image sequence is processed to obtain the distribution probability of the fire fighting potential safety hazard corresponding to each position pixel point, and the distribution probability of the fire fighting potential safety hazard of the pixel point at the position (x, y) is w3(x, y); the greater the distribution probability value of the fire safety hazard, the greater the probability that the corresponding position has the dangerous goods, the more the position where the dangerous goods are located needs to be vigilant, and therefore the more the corresponding position needs to be illuminated by a fire emergency illuminating lamp to avoid the approach of people.
Thirdly, performing feature fusion according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to each position to obtain the fire emergency lighting probability of each position pixel point; the calculation formula for obtaining the probability of fire emergency lighting at the position (x, y) in this embodiment is:
wherein, w (x, y) is the fire emergency lighting probability value of position (x, y) department, and the value range of w (x, y) is [0,1], and the value of personnel distribution probability, fire safety hidden danger distribution probability and fire equipment distribution probability is big more in this embodiment, and then fire emergency lighting probability value is just big more, and fire emergency lighting probability value is big more, then indicates that the corresponding position needs to be lighted by fire emergency lighting lamp more. As other implementation modes, different weights can be set for the personnel distribution probability, the fire safety hazard distribution probability and the fire fighting equipment distribution probability, and then the fire emergency lighting probability corresponding to each position is obtained in a weighted summation mode.
The fire emergency lighting probability corresponding to each position pixel point is obtained according to the formula, and the pixel value corresponding to each position pixel point is set to be the corresponding fire emergency lighting probability value, so that a fire emergency lighting probability graph corresponding to the global image is obtained.
And S3, inputting the fire emergency lighting probability map and the quantity of the fire emergency lighting lamps into the trained fire emergency lighting lamp layout optimization network, and generating a corresponding fire emergency lighting lamp layout scheme.
In the embodiment, a plurality of rooms in the target building space are considered, so that the global image is divided into a plurality of sub-areas according to different rooms in the target building space, and a fire emergency lighting probability map corresponding to each sub-area is obtained according to the fire emergency lighting probability map corresponding to the global image and is recorded as a sub-fire emergency lighting probability map. Because the purposes of different rooms are different, the number of the fire-fighting emergency illuminating lamps required by different rooms is different, and the implementation acquires the number of the fire-fighting emergency illuminating lamps required by each subarea according to the purposes of the rooms. The neutron regions in this embodiment may be: office areas, electrical distribution rooms, fire control rooms, and other areas of use, and the number of fire emergency lighting lamps that need to be installed in different areas is known.
In order to obtain the layout scheme of the fire-fighting emergency lamps of all the subregions in the global image and optimize the layout of the fire-fighting emergency lamps of all the rooms in the current target building space, a fire-fighting emergency lamp layout optimization network is constructed, and then the layout scheme of the fire-fighting emergency lamps of all the rooms in the current target building space can be generated by means of the fire-fighting emergency lamp layout optimization network. The method for constructing the layout optimization network of the fire-fighting emergency illuminating lamp comprises the following steps:
the emergency lighting lamp layout optimization network in the implementation comprises a generator sub-network and a discriminator sub-network, wherein:
the generator sub-network is of an Encoder1-Decoder1 structure, and a sub-fire emergency lighting probability graph corresponding to a sub-region and the number of fire emergency lighting lamps required in the sub-region are input; the Encoder1 is used for carrying out downsampling operation on an input image to obtain a feature map corresponding to the input image, then the feature map is leveled to obtain a one-dimensional vector, the obtained one-dimensional vector and the number of fire emergency lamps are subjected to concatemate operation and then are sent into the Decoder1, and the Decoder1 is used for obtaining an emergency fire-fighting illumination area in a sub-area. The emergency fire-fighting lighting area comprises a plurality of optimal lighting areas in the area, and the number of the lighting areas is the same as that of the input fire-fighting emergency lighting lamps.
The discriminator sub-network is of an Encoder2+2 xFC structure, a plurality of lighting areas generated by the generator sub-network are input, the discriminator sub-network inputs the input lighting areas into the first classifier FC1, the judgment result of whether each lighting area meets the real condition or not is output, and the first classifier is used for restricting the plurality of lighting areas generated by the generator sub-network to meet the actual condition; when the determination result is true, the lighting area that matches the reality is input to the second classifier FC2, and the position information of the fire emergency lighting lamp corresponding to the lighting area is obtained.
The training process of the fire-fighting emergency lighting lamp layout optimization network in the embodiment is as follows:
first, the present embodiment first trains a sub-network of discriminators, and a training set for training the sub-network of discriminators is a training set of illumination areas, which includes actual illumination areas of fire emergency lamps at different locations and randomly generated illumination areas. The actual lighting area is directly obtained through a simulator and is obtained by simulating the lighting area formed by the fire-fighting emergency lighting lamps at different positions on the ground, and the lighting area accords with the actual situation; the randomly generated illumination region is an illumination region of an arbitrary shape, and is not suitable for practical use. The position information of the fire suppression emergency lamp in the embodiment is expressed by coordinates in a three-dimensional space, and is marked as (x, y, z).
Then, corresponding first label data are set for each illumination area in the illumination area training set, the illumination area which is in line with the actual condition in the first label data is marked as 1, the illumination area which is not in line with the actual condition is marked as 0, namely the first label data are divided into two types which are in line with the actual condition and are not in line with the actual condition, and the first label data are label data of a first classifier; in the embodiment, the position information of the fire-fighting emergency lighting lamp corresponding to the lighting area, of which the first label data accords with the actual lighting area, is recorded as the actual position information; the position information of the fire-fighting emergency illuminating lamp corresponding to the illumination area with the first label data being not in line with the actual illumination area is set to be (0,0,0), the position information is recorded as the position information not in line with the actual illumination area, then the obtained actual position information and the position information not in line with the actual illumination area are used as second label data, and the second label data are used for training a second classifier. The embodiment trains a discriminator sub-network by using the illumination area training set, the first label data and the second label data to obtain the trained discriminator sub-network. The penalty function for the arbiter subnetwork is calculated as:
Loss1＝LFC1+LFC2
therein, Loss1For the value of the loss function of the discriminator subnetwork, LFC1For the loss function value of the first classifier, the first classifier in this embodiment uses a two-class cross entropy loss function, LFC2Num represents the number of samples of a batch of the training set of illumination areas for the loss function value of the second classifier;
the first label data of the ith training sample,
is the judgment result (0 or 1) output by the first classifier,
the second label data for the ith training sample,
is the result output by the second classifier. The embodiment continuously updates the network parameters by using a gradient descent method, thereby completing the training of the discriminator subnetwork.
The loss function of the second classifier is used for obtaining accurate position information of the fire-fighting emergency lighting lamp corresponding to the lighting area, and when the difference between the output position information of the fire-fighting emergency lighting lamp corresponding to the lighting area and the corresponding second label data is smaller, the result output by the network is more accurate.
Second, the present embodiment freezes the network parameters of the trained discriminator subnetwork, and trains the generator subnetwork. The training set of the generator subnetwork is a historical fire emergency lighting probability map, and then the generator subnetwork is trained by means of the discriminator subnetwork. In this embodiment, when the lighting areas generated by the generator subnetworks do not overlap with each other, the lighting in the largest range can be realized, and when the sum of the probabilities of the fire emergency lighting in the lighting areas is larger and the position with the largest probability is closer to the center of the area, the lighting effect of the fire emergency lighting is better. The formula for the loss function of the generator subnetwork is:
therein, Loss2For generating loss function values of a sub-network of devices, ROI11 st illumination area, ROI, generated for the generator subnetwork22 nd illumination region, ROI, generated for a generator subnetworkNNth illumination area, ROI, generated for a generator subnetworkfTo generate the f-th illumination area for the generator sub-network,
as the abscissa of the center point of the f-th illumination area,
is the ordinate of the center point of the f-th illumination area,
is the lighting probability corresponding to the center point of the f-th lighting area, w (x, y) is the lighting probability of the pixel point with x as the abscissa and y as the ordinate in the f-th lighting area,
the result of the discrimination corresponding to the f-th illumination region output by the first classifier of the discriminator sub-network is N, the number of illumination regions generated by the generator sub-network, and e is a natural constant.
The first part (ROI) of the loss function of the generator subnetwork1∩ROI2∩…∩ROIN) For ensuring that the generated plurality of illumination areas do not overlap with each other; the second part
The device is used for ensuring that the position with the maximum probability of the sum of the fire emergency lighting probabilities in the lighting area is close to the center; third part
Is to ensure multiple illumination areas generatedAccording with the actual lighting situation. In this embodiment, the generator subnetwork uses the gradient descent method to update the network parameters to complete the training of the generator subnetwork.
After the training of the fire-fighting emergency lighting lamp layout optimization network is completed, the implementation firstly inputs the sub fire-fighting emergency lighting probability maps corresponding to the sub areas and the quantity of the fire-fighting emergency lighting lamps corresponding to the sub areas into the generator sub-network of the trained fire-fighting emergency lighting lamp layout optimization network to obtain a plurality of lighting areas consistent with the quantity of the fire-fighting emergency lighting lamps in the sub areas, then inputs the generated lighting areas into the discriminator sub-network, and the FC2 of the discriminator sub-network directly obtains the position information of the fire-fighting emergency lighting lamps corresponding to the lighting areas, thereby obtaining the fire-fighting emergency lighting lamp layout scheme of the sub areas.
And step S4, adjusting the original layout of the corresponding fire-fighting emergency lighting lamps in the building information model according to the fire-fighting emergency lighting lamp layout scheme.
According to the step S3, the layout scheme of the fire-fighting emergency illuminating lamps corresponding to each subarea is obtained, so that the target position information of each fire-fighting emergency illuminating lamp in each subarea can be obtained according to the layout scheme of the fire-fighting emergency illuminating lamps corresponding to each subarea, namely the positions of each fire-fighting emergency illuminating lamp in the generated layout scheme of the fire-fighting emergency illuminating lamps; then obtaining the position information of the current fire-fighting emergency illuminating lamp through a building information model and recording the position information as original position information, namely original layout; in this embodiment, the original location information of each fire emergency lighting lamp in each subarea is recorded as Z ═ Z1,z2,…,zNRecording the target position information of each fire-fighting emergency lighting lamp as
In the embodiment, each original position information and each target position information are regarded as nodes, and a bipartite graph corresponding to each sub-region is constructed, wherein the L2 distance between each original position information node and each target position information node is taken as the edge weight of each edge in the bipartite graph, wherein the node zuAnd node
Side weight p ofu,vThe calculation method comprises the following steps:
wherein z isuFor each of the u-th original position information,
for the v-th object position information, zuAnd
z is greater the distance L2uAnd
the smaller the weight of the edge between the two, i.e. the greater the difference between the two, i.e. the less likely it is to be called a matching pair in the subsequent pairing process.
In order to determine whether the original layout of the fire emergency lighting lamps in each sub-area needs to be adjusted, in this embodiment, first, matching pairs corresponding to the original position information and the target position information of each fire emergency lighting lamp in the sub-area one to one are obtained, the matching pairs form a matching pair set corresponding to the sub-area, then, the reasonable degree of the original layout of each fire emergency lighting lamp in the subsequent sub-area is analyzed, and the determination is specifically performed according to the reasonable degree corresponding to each sub-area:
processing the bipartite graphs corresponding to the sub-regions by using a KM algorithm to obtain a matching pair set of original position information and target position information of each fire-fighting emergency lighting lamp in each sub-region, wherein the matching pair set is an optimal set with the edge weight value between two position nodes in each matching pair being the maximum, namely the distance between the two position nodes in each matching pair is the minimum; then, according to the matching pair set corresponding to each subregion, calculating the reasonable degree of the original layout of the fire-fighting emergency lighting lamps corresponding to each subregion, comparing the original position information with the corresponding target position information by the reasonable degree of the original layout of the fire-fighting emergency lighting lamps corresponding to each subregion, if the difference between the original position information and the corresponding target position information is larger, the smaller the value of the corresponding reasonable degree is, the more the position of the fire-fighting emergency lighting lamps of the subregion needs to be adjusted, and the calculation formula of the reasonable degree of the original layout of the fire-fighting emergency lighting lamps in each subregion is as follows:
wherein, for the reasonable degree of the original layout of the fire emergency lighting lamps in the Score subregion, L2(j) is the L2 distance between the original position information and the target position information in the jth matching pair, and M is the total number of the matching pairs in the matching pair set corresponding to the subregion. In this embodiment, the value range of the Score is [0,1], and a larger Score value indicates that the current original layout is more reasonable, so that the adjustment is more unnecessary.
According to the method, whether the original layout of the fire-fighting emergency lighting lamps corresponding to the sub-areas needs to be adjusted or not is judged according to the obtained reasonable degree of the original layout of the fire-fighting emergency lighting lamps corresponding to the sub-areas, and when the reasonable degree corresponding to the sub-areas is smaller than a reasonable degree threshold value, the situation that the original layout of the emergency fire-fighting lighting lamps is greatly different from a generated fire-fighting emergency lighting lamp layout scheme is shown, and the original layout of the fire-fighting lighting lamps cannot meet the requirements of an actual environment, so that the sub-areas with the reasonable degree larger than the reasonable degree threshold value are integrated according to corresponding matching pairs to adjust the original positions of the fire-fighting emergency lighting lamps to corresponding target positions, so that the fire-fighting emergency lamps can meet the actual requirements when a fire disaster occurs, and the safe evacuation of personnel and the smooth proceeding of fire-fighting operation are guaranteed; when the reasonable degree corresponding to the sub-area is greater than or equal to the reasonable degree threshold value, the original layout of the current fire-fighting emergency illuminating lamp can meet the requirement of the actual environment, and therefore the original layout of the fire-fighting emergency illuminating lamp in the corresponding sub-area does not need to be adjusted. The reasonable degree threshold value is set to be 0.6 in the embodiment, and the reasonable degree threshold value can be set according to actual needs in the specific implementation process.
According to the method, the obtained building information model in the target building space and the global image of the target building space are analyzed, then the corresponding fire emergency lighting probability map is obtained according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to each position pixel point, finally, the fire emergency lighting probability map and the quantity of the fire emergency lighting lamps are input into the trained fire emergency lighting lamp layout optimization network, and a fire emergency lighting lamp layout scheme for adjusting the original layout of the fire emergency lighting lamps is generated. The invention optimizes the layout of the fire-fighting emergency lighting lamps in the target building space by utilizing the fire-fighting emergency lighting lamp layout optimization network so as to ensure that the layout of the fire-fighting emergency lighting lamps can meet the actual requirement.
It should be noted that: the above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (7)
1. A fire-fighting emergency illuminating lamp layout optimization method based on artificial intelligence is characterized by comprising the following steps:
acquiring a building information model of a target building space and a global image of the target building space;
obtaining a corresponding fire emergency lighting probability map according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire equipment distribution probability corresponding to each pixel point in the global image;
inputting the fire emergency lighting probability map and the number of fire emergency lighting lamps into a trained fire emergency lighting lamp layout optimization network to generate a corresponding fire emergency lighting lamp layout scheme;
and adjusting the original layout of the corresponding fire-fighting emergency lighting lamps in the building information model according to the fire-fighting emergency lighting lamp layout scheme.
2. The artificial intelligence based fire emergency lighting lamp layout optimization method according to claim 1, wherein the fire emergency lighting lamp layout optimization network comprises a generator sub-network and a discriminator sub-network, and the training of the fire emergency lighting lamp layout optimization network comprises:
acquiring a lighting area training set, wherein the lighting area training set comprises actual lighting areas of fire-fighting emergency lighting lamps at different positions and randomly generated lighting areas; setting corresponding first label data for each lighting area in a lighting area training set, wherein the first label data comprises: actual and non-actual compliance;
recording the lighting area in which the first label data accords with the reality as a first lighting area, and taking the position information of the fire-fighting emergency lighting lamp corresponding to the first lighting area as second label data of the first lighting area; recording the illumination area in which the first label data does not conform to the reality as a second illumination area, and using the set position information as second label data of the second illumination area;
training a discriminator sub-network according to a lighting area training set, first label data and second label data, wherein the first label data are label data of a first classifier of the discriminator sub-network, the second label data are label data of a second classifier of the discriminator sub-network, and the discriminator sub-network is used for obtaining position information of a fire-fighting emergency lighting lamp corresponding to a lighting area;
freezing the network parameters of the trained discriminator sub-network, and training a generator sub-network by using the trained discriminator sub-network and a historical fire emergency lighting probability map, wherein the generator sub-network is used for generating a lighting area according with the actual situation.
3. A fire emergency lighting lamp layout optimization method based on artificial intelligence as claimed in claim 2, wherein the calculation formula of the loss function of the generator sub-network is:
therein, Loss2For generating loss function values of a sub-network of devices, ROI11 st illumination area, ROI, generated for the generator subnetwork22 nd illumination region, ROI, generated for a generator subnetworkNNth illumination area, ROI, generated for a generator subnetworkfTo generate the f-th illumination area for the generator sub-network,
as the abscissa of the center point of the f-th illumination area,
is the ordinate of the center point of the f-th illumination area,
is the lighting probability corresponding to the center point of the f-th lighting area, w (x, y) is the lighting probability of the pixel point with x as the abscissa and y as the ordinate in the f-th lighting area,
the result of the discrimination corresponding to the f-th illumination region output by the first classifier of the discriminator sub-network is N, the number of illumination regions generated by the generator sub-network, and e is a natural constant.
4. The method according to claim 1, wherein the step of inputting the probability map and the number of the fire emergency lamps into the trained fire emergency lamp layout optimization network comprises the steps of:
dividing the global image into a plurality of sub-regions according to the positions of all rooms in the building information model;
dividing the fire emergency lighting probability map according to each corresponding subregion in the global image to obtain a sub fire emergency lighting probability map corresponding to each subregion;
and inputting the sub-fire emergency lighting probability map corresponding to each subregion in the global image and the number of the fire emergency lighting lamps corresponding to each subregion into a trained fire emergency lighting lamp layout optimization network to generate a fire emergency lighting lamp layout scheme corresponding to each subregion.
5. The method according to claim 4, wherein adjusting the layout of the corresponding fire-fighting emergency lighting lamps in the building information model according to the layout scheme of the fire-fighting emergency lighting lamps comprises:
acquiring original position information of each fire-fighting emergency lighting lamp in each subregion according to the layout information of the fire-fighting emergency lighting lamps in the building information model;
constructing a corresponding bipartite graph according to the original position information of each fire-fighting emergency lighting lamp in each subarea and the target position information of each fire-fighting emergency lighting lamp corresponding to the fire-fighting emergency lighting lamp layout scheme corresponding to each subarea;
processing bipartite graphs corresponding to the sub-regions by using a KM algorithm to obtain a matching pair set of original position information and target position information of the fire-fighting emergency lighting lamps in the sub-regions;
and calculating the reasonable degree of the original layout of the fire-fighting emergency illuminating lamps corresponding to the sub-regions according to the matching pair set corresponding to the sub-regions, judging whether the reasonable degree of the original layout of the fire-fighting emergency illuminating lamps corresponding to the sub-regions is smaller than a reasonable degree threshold value, and if so, adjusting the original positions of the fire-fighting emergency illuminating lamps in the corresponding sub-regions to the corresponding target positions.
6. The method for optimizing the layout of the fire-fighting emergency lighting lamp based on the artificial intelligence as claimed in claim 1, wherein the method for obtaining the corresponding probability map of the fire-fighting emergency lighting lamp according to the personnel distribution probability, the fire-fighting safety hidden danger degree probability and the fire-fighting equipment distribution probability corresponding to each pixel point of the global image comprises the following steps:
acquiring continuous multi-frame global images, and recording as a global image sequence;
carrying out binarization processing on each image according to the position of a personnel key point corresponding to each image in the global image sequence to obtain a personnel distribution diagram sequence; calculating the average value of the pixel values of all pixel points corresponding to all positions in the personnel distribution map sequence, and recording the average value as the personnel distribution probability corresponding to the pixel points of all positions;
inputting the global image sequence into a semantic segmentation network to obtain a fire fighting equipment mask image and a dangerous goods mask image;
carrying out binary processing on each image in the global image sequence according to the fire fighting equipment shade image to obtain a fire fighting equipment shade binary image sequence; calculating the average value of pixel values of all pixel points corresponding to all positions in the fire fighting equipment shade binary image sequence, and recording the average value as the fire fighting equipment distribution probability corresponding to the pixel points of all the positions;
carrying out binary processing on each image in the global image sequence according to the dangerous article mask image to obtain a dangerous article mask binary image sequence; calculating the average value of the pixel values of all pixel points corresponding to all positions in the binary image sequence of the hazardous article shade, and recording the average value as the distribution probability of the fire safety hidden danger corresponding to the pixel points of all the positions;
obtaining the lighting probability corresponding to each position pixel point according to the personnel distribution probability, the fire safety hidden danger distribution probability and the fire fighting equipment distribution probability corresponding to each position pixel point;
and taking the illumination probability corresponding to each position pixel point as the pixel value of the corresponding position pixel point to obtain a fire-fighting emergency illumination probability map.
7. The artificial intelligence-based fire-fighting emergency lighting lamp layout optimization method according to claim 5, wherein the calculation formula for calculating the reasonable degree of the original fire-fighting emergency lighting lamp layout corresponding to each sub-area is as follows:
wherein, Score is the reasonable degree of original layout of fire emergency lighting lamps in the sub-region, L2(j) is the L2 distance between the original position information and the target position information in the jth matching pair, and M is the total number of matching pairs in the matching pair set corresponding to the sub-region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111507302.4A CN114186735B (en) | 2021-12-10 | 2021-12-10 | Fire emergency lighting lamp layout optimization method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111507302.4A CN114186735B (en) | 2021-12-10 | 2021-12-10 | Fire emergency lighting lamp layout optimization method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114186735A true CN114186735A (en) | 2022-03-15 |
| CN114186735B CN114186735B (en) | 2023-10-20 |
Family
ID=80604383
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111507302.4A Active CN114186735B (en) | 2021-12-10 | 2021-12-10 | Fire emergency lighting lamp layout optimization method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114186735B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115310882A (en) * | 2022-10-12 | 2022-11-08 | 北京晨豪科技有限公司 | Visual fire-fighting strategy planning method and system |
| CN117094230A (en) * | 2023-10-17 | 2023-11-21 | 中建照明有限公司 | Urban lamplight optimization model training system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202050543U (en) * | 2011-01-13 | 2011-11-23 | 广东和新科技有限公司 | Emergency communication lighting terminal device based on technology of Internet of things |
| CN109185743A (en) * | 2018-08-28 | 2019-01-11 | 宁波飞拓电器有限公司 | A kind of intelligent fire emergency-lighting system |
| CN109978807A (en) * | 2019-04-01 | 2019-07-05 | 西北工业大学 | A kind of shadow removal method based on production confrontation network |
| CN110046837A (en) * | 2019-05-20 | 2019-07-23 | 北京唐芯物联网科技有限公司 | A kind of fire management system based on artificial intelligence |
| WO2021017372A1 (en) * | 2019-08-01 | 2021-02-04 | 中国科学院深圳先进技术研究院 | Medical image segmentation method and system based on generative adversarial network, and electronic equipment |
-
2021
- 2021-12-10 CN CN202111507302.4A patent/CN114186735B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202050543U (en) * | 2011-01-13 | 2011-11-23 | 广东和新科技有限公司 | Emergency communication lighting terminal device based on technology of Internet of things |
| CN109185743A (en) * | 2018-08-28 | 2019-01-11 | 宁波飞拓电器有限公司 | A kind of intelligent fire emergency-lighting system |
| CN109978807A (en) * | 2019-04-01 | 2019-07-05 | 西北工业大学 | A kind of shadow removal method based on production confrontation network |
| CN110046837A (en) * | 2019-05-20 | 2019-07-23 | 北京唐芯物联网科技有限公司 | A kind of fire management system based on artificial intelligence |
| WO2021017372A1 (en) * | 2019-08-01 | 2021-02-04 | 中国科学院深圳先进技术研究院 | Medical image segmentation method and system based on generative adversarial network, and electronic equipment |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115310882A (en) * | 2022-10-12 | 2022-11-08 | 北京晨豪科技有限公司 | Visual fire-fighting strategy planning method and system |
| CN115310882B (en) * | 2022-10-12 | 2023-01-24 | 北京晨豪科技有限公司 | Visual fire-fighting strategy planning method and system |
| CN117094230A (en) * | 2023-10-17 | 2023-11-21 | 中建照明有限公司 | Urban lamplight optimization model training system and method |
| CN117094230B (en) * | 2023-10-17 | 2024-02-13 | 中建照明有限公司 | Urban lamplight optimization model training system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114186735B (en) | 2023-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bu et al. | Intelligent and vision-based fire detection systems: A survey | |
| CN114186735A (en) | Fire-fighting emergency illuminating lamp layout optimization method based on artificial intelligence | |
| Cho et al. | A neural-based crowd estimation by hybrid global learning algorithm | |
| Kolesenkov et al. | Anthropogenic situation express monitoring on the base of the fuzzy neural networks | |
| CN108389359B (en) | Deep learning-based urban fire alarm method | |
| CN107729799A (en) | Crowd's abnormal behaviour vision-based detection and analyzing and alarming system based on depth convolutional neural networks | |
| Khalil et al. | Fire detection using multi color space and background modeling | |
| CN110414400A (en) | A kind of construction site safety cap wearing automatic testing method and system | |
| CN104320620A (en) | Video monitoring and network voice integrated urban emergency management system | |
| CN107688793A (en) | A kind of outside transformer substation fire automatic monitoring method for early warning | |
| CN116664359A (en) | Intelligent fire early warning decision system and method based on multi-sensor fusion | |
| CN114689058A (en) | Fire evacuation path planning method based on deep learning and hybrid genetic algorithm | |
| Dung et al. | Algorithm for fire detection using a camera surveillance system | |
| CN110503017A (en) | Wisdom energy conservation occupancy detection system and method based on image procossing | |
| CN114120171A (en) | Fire smoke detection method, device and equipment based on video frame and storage medium | |
| CN116189372A (en) | Forest fire prevention multilayer early warning and monitoring system and method applying big data technology | |
| CN116437538A (en) | Dimming method and system for crowd gathering monitoring based on multifunctional intelligent lamp post | |
| CN114821486B (en) | Personnel identification method in power operation scene | |
| KR102432675B1 (en) | Monitoring and safety alarm system for abnormal signs and drowning accidents of pool users using CCTV images | |
| KR20210047490A (en) | Fire risk predication system using unmanned aerial vehicle and method thereof | |
| CN113569801B (en) | Distribution construction site live equipment and live area identification method and device thereof | |
| Iqbal et al. | Deep learning algorithm for fire detection | |
| KR102305955B1 (en) | Emergency guide lamp comprising counting function and access personnel management method using the same | |
| CN112699745A (en) | Method for positioning trapped people on fire scene | |
| Chen | A fuzzy-based approach for smart building evacuation modeling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |