CN112036279A - Intelligent building monitoring method and system - Google Patents
Intelligent building monitoring method and system Download PDFInfo
- Publication number
- CN112036279A CN112036279A CN202010853100.4A CN202010853100A CN112036279A CN 112036279 A CN112036279 A CN 112036279A CN 202010853100 A CN202010853100 A CN 202010853100A CN 112036279 A CN112036279 A CN 112036279A
- Authority
- CN
- China
- Prior art keywords
- image
- smoking
- target monitoring
- characteristic
- feature
- 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.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 201
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000000391 smoking effect Effects 0.000 claims abstract description 279
- 230000006399 behavior Effects 0.000 claims abstract description 123
- 230000001537 neural effect Effects 0.000 claims abstract description 29
- 235000019504 cigarettes Nutrition 0.000 claims description 99
- 230000009467 reduction Effects 0.000 claims description 30
- 238000013528 artificial neural network Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 17
- 238000011176 pooling Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 4
- 239000000779 smoke Substances 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 19
- 238000004590 computer program Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 201000004792 malaria Diseases 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- GOLXNESZZPUPJE-UHFFFAOYSA-N spiromesifen Chemical compound CC1=CC(C)=CC(C)=C1C(C(O1)=O)=C(OC(=O)CC(C)(C)C)C11CCCC1 GOLXNESZZPUPJE-UHFFFAOYSA-N 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06N3/045—Combinations of 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
Abstract
The invention relates to a building intelligent monitoring method and a system, which comprises the steps of obtaining a target monitoring image of a target floor; inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image; according to the feature image of the target monitoring image, smoking feature similarity in the feature image is calculated, and the smoking feature similarity is used as smoking behavior probability; and if the smoking behavior probability is greater than the preset threshold probability, generating smoking warning information, and sending the smoking warning information to a monitoring terminal associated with smoking. The invention can automatically monitor and alarm the smoking behavior of smokers, thereby improving the environmental quality of floors.
Description
Technical Field
The invention relates to the technical field of building monitoring, in particular to a building intelligent monitoring method and system.
Background
At present, monitoring equipment is installed in office buildings, hospitals, factories and other places, and is used for intelligent security driving and protecting, but most of the existing building monitoring systems are used for monitoring suspicious behaviors such as theft or damage, and cannot be used for monitoring special behaviors, such as smoking behaviors.
Although smoking is prohibited or even fine is paid in many occasions, the smoking behavior of people is not completely eradicated, so the inventor thinks that under the condition of no related monitoring constraint, smoking behavior can cause the malaria in the corridor, seriously affect the air quality of the floor and even have the risk of fire.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides an intelligent building monitoring method and system, which can automatically monitor and alarm the smoking behavior of smokers, thereby improving the environmental quality of floors.
The above object of the present invention is achieved by the following technical solutions:
an intelligent building monitoring method comprises the following steps:
acquiring a target monitoring image of a target floor;
inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
according to the feature image of the target monitoring image, smoking feature similarity in the feature image is calculated, and the smoking feature similarity is used as smoking behavior probability;
and if the smoking behavior probability is greater than the preset threshold probability, generating smoking warning information, and sending the smoking warning information to a monitoring terminal associated with smoking.
By adopting the technical scheme, if a person smokes on a floor or a staircase, the target monitoring image is obtained through the monitoring camera, the smoking behavior probability existing in the target monitoring image can be identified and judged through the preset smoking identification deep neural model, and when the smoking behavior probability is greater than the preset threshold probability, smoking alarm information can be sent to the associated monitoring end, so that the current area of the person smoking can be reminded to forbid smoking, smoking behaviors are reduced, and the air quality of the floor is improved; the target monitoring image is input into a preset smoking identification deep neural model, the characteristic image of the target monitoring image is automatically extracted, and the similarity with smoking behaviors is calculated from the characteristic image, so that the probability of smoking behaviors existing in the target monitoring image can be determined, the smoking behaviors are intelligently identified, and the accuracy of smoking behavior identification is improved.
Optionally, the step of obtaining the target monitoring image of the target floor includes:
the method comprises the steps of obtaining a target monitoring video of a target floor, decomposing the target monitoring video into a single-frame image of a time sequence through OpenCV software, and taking the single-frame image as the target monitoring image.
By adopting the technical scheme, the target monitoring video is subjected to framing through the OpenCV software to obtain the single-frame images of the time sequence, so that the subsequent model can perform characteristic analysis on smoking behaviors in the single-frame images and can perform real-time analysis.
Optionally, before the step of inputting the target monitoring image into the preset smoking recognition deep neural model in time sequence, the method further includes:
and carrying out normalization processing and data enhancement processing on the target monitoring image.
By adopting the technical scheme, before the target monitoring image is input into the smoking identification deep neural model, the target monitoring image is normalized, so that the accuracy of subsequent image identification can be improved, the overfitting of the subsequent image identification can be reduced by performing data enhancement processing on the target monitoring image, and the identification error rate can be reduced.
Optionally, the preset smoking recognition deep neural model comprises a cigarette recognition network, a hand-held cigarette recognition network and a mouth smoking recognition network; the characteristic images comprise a first characteristic image, a second characteristic image and a third characteristic image; the step of inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image comprises the following steps:
acquiring a first characteristic image of a target monitoring image through the cigarette identification network;
acquiring a second characteristic image of a target monitoring image through the hand-held cigarette identification network;
and acquiring a third characteristic image of the target monitoring image through the mouth smoking identification network.
By adopting the technical scheme, the preset smoking identification deep neural model comprises a cigarette identification network, a hand-held cigarette identification network and a mouth smoking identification network, and the three networks are used for identifying the target monitoring image respectively to obtain the corresponding characteristic image, so that the local characteristics related to smoking in the target monitoring image can be extracted at multiple angles.
Optionally, the step of obtaining the first characteristic image of the target monitoring image through the cigarette identification network includes:
the cigarette identification network obtains an original cigarette characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original cigarette characteristic image through a network pooling layer, and takes an image corresponding to the data dimension reduction as a first characteristic image of the target monitoring image.
By adopting the technical scheme, the cigarette identification network can obtain rich original cigarette characteristic images through the convolution layers of the channels, data dimension reduction is carried out on the original cigarette characteristic images through the network pooling layer, the data amount can be reduced, and then the corresponding images after the data dimension reduction are used as the first characteristic images of the target monitoring images.
Optionally, the step of obtaining a second feature image of the target monitoring image through the hand-held cigarette recognition network includes:
the hand-held cigarette identification network obtains an original hand-held cigarette characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original hand-held cigarette characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a second characteristic image of the target monitoring image.
Through the technical scheme, the hand-held cigarette identification network can acquire rich original hand-held cigarette characteristic images through the convolution layers of the channels, performs data dimension reduction on the original hand-held cigarette characteristic images through the network pooling layer, can reduce data volume, and then takes the corresponding images after the data dimension reduction as the second characteristic images of the target monitoring images.
Optionally, the step of obtaining a third feature image of the target monitoring image through the mouth smoking recognition network includes:
the mouth smoking identification network obtains an original mouth smoking characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original mouth smoking characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a third characteristic image of the target monitoring image.
According to the technical scheme, the mouth smoking identification network can acquire rich original mouth smoking characteristic images through the convolution layers of the channels, data dimension reduction is carried out on the original mouth smoking characteristic images through the network pooling layer, the data quantity can be reduced, and then the corresponding images after the data dimension reduction are used as third characteristic images of the target monitoring images.
Optionally, the calculating, according to the feature image of the target monitoring image, smoking feature similarity in the feature image, and taking the smoking feature similarity as smoking behavior probability includes:
the cigarette identification network outputs the similarity containing the cigarette characteristic information in the first characteristic image through a radial basis function neural network layer, and the similarity containing the cigarette characteristic information is used as the probability of the first smoking behavior;
the hand-held cigarette identification network outputs the similarity containing the hand-held cigarette characteristic information in the second characteristic image through the radial basis function neural network layer, and takes the similarity containing the hand-held cigarette characteristic information as a second smoking behavior probability;
and the mouth smoking identification network outputs the similarity containing the mouth smoking characteristic information in the third characteristic image through the radial basis function neural network layer, and takes the similarity containing the mouth smoking characteristic information as a third smoking behavior probability.
Through the technical scheme, the corresponding feature similarity is respectively output through the cigarette identification network, the hand-held cigarette identification network and the hand-held cigarette identification network, so that the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability can be accurately determined, the action occurrence probability related to smoking can be analyzed at multiple angles, and the intelligent monitoring on the smoking behavior is realized.
Optionally, the calculating, according to the feature image of the target monitoring image, smoking feature similarity in the feature image, and taking the smoking feature similarity as smoking behavior probability further includes:
and setting the first smoking behavior probability as a first weight, setting the second smoking behavior probability as a second weight, setting the third smoking behavior probability as a third weight, and multiplying the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability by probability values obtained by corresponding weights respectively to serve as the smoking behavior probabilities.
Through the technical scheme, different weights are respectively set for the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability, so that the three smoking behavior probabilities can be comprehensively considered, and the accuracy of the finally analyzed smoking behavior probability is improved.
The second aim of the invention is realized by the following technical scheme:
an intelligent building monitoring system comprising:
the acquisition module is used for acquiring a target monitoring image of a target floor;
the characteristic extraction module is used for inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
the probability calculation module is used for calculating smoking characteristic similarity in the characteristic image according to the characteristic image of the target monitoring image and taking the smoking characteristic similarity as smoking behavior probability;
and the judgment warning module is used for generating smoking warning information if the smoking behavior probability is greater than a preset threshold probability, and sending the smoking warning information to a monitoring terminal associated with smoking.
According to the technical scheme, the feature image of the target monitoring image can be rapidly extracted through the feature extraction module, the similarity with smoking feature information in the feature image can be calculated through the probability calculation module, so that the probability of smoking behavior in the target monitoring image can be automatically identified, and then the smoking warning information is sent to the associated monitoring terminal when the probability of smoking behavior is judged to be greater than the threshold probability through the judgment warning module, so that the effectiveness of floor smoking prohibition management and the floor environment quality are improved.
In summary, the invention includes at least one of the following beneficial technical effects:
1. the smoking behavior probability existing in the target monitoring image can be identified and judged through the preset smoking identification deep neural model, and when the smoking behavior probability is larger than the preset threshold probability, smoking warning information can be sent to the associated monitoring end, so that a person who is smoking can be reminded that smoking is forbidden in the current area, smoking behaviors are reduced, and the air quality of a floor is improved.
2. The target monitoring image is input into a preset smoking identification deep neural model, the characteristic image of the target monitoring image is automatically extracted, and the similarity with smoking behaviors is calculated from the characteristic image, so that the probability of smoking behaviors existing in the target monitoring image can be determined, the smoking behaviors are intelligently identified, and the accuracy of smoking behavior identification is improved.
3. The preset smoking identification deep neural model comprises a cigarette identification network, a hand-held cigarette identification network and a mouth smoking identification network, and the three networks are used for respectively identifying the target monitoring image to obtain corresponding characteristic images, so that local characteristics related to smoking in the target monitoring image can be extracted at multiple angles; different weights are set for the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability respectively, so that the three smoking behavior probabilities can be comprehensively considered, and the accuracy of the finally analyzed smoking behavior probability is improved.
Drawings
FIG. 1 is a flow chart of an implementation of an intelligent building monitoring method according to an embodiment of the present application;
FIG. 2 is a flowchart illustrating an implementation of step S20 according to an embodiment of the present application;
FIG. 3 is a flowchart illustrating an implementation of step S30 according to an embodiment of the present application;
FIG. 4 is a functional block diagram of an intelligent building monitoring system according to an embodiment of the present application;
FIG. 5 is a functional block diagram of a computer device according to an embodiment of the present application.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
Example (b):
in this embodiment, as shown in fig. 1, the present application discloses a building intelligent monitoring method, which includes the following steps:
s10: and acquiring a target monitoring image of a target floor.
In this embodiment, the target floor refers to a floor to be monitored of a building to be monitored; the target monitoring image is a picture image for monitoring a certain area of a target floor.
Specifically, a target monitoring video is shot through a monitoring camera, and then a target monitoring image is obtained from the target monitoring video.
Further, the target monitoring video is decomposed into a time sequence of single-frame images through OpenCV software, and the single-frame images are used as the target monitoring images. In the present embodiment, since the number of image frames included in the target surveillance video is excessive, it is set to rotate the image once every 30 frames.
In this embodiment, normalization processing may be performed on the target monitoring image, and the target monitoring image is processed into a three-channel image with 224 × 224 pixels; and simultaneously, performing data enhancement processing on the target monitoring object, including horizontal mirror image turning and noise increasing processing.
S20: and inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image.
In the present embodiment, the feature image is an image formed by extracting local features from the target monitor image by the smoking recognition deep neural model.
Specifically, the smoking identification deep neural model comprises a cigarette identification network, a hand-held cigarette identification network and a mouth smoking identification network, the cigarette identification network, the hand-held cigarette identification network and the mouth smoking identification network adopt a convolutional neural network AlexNet architecture, a large number of smoking behavior images are obtained from the Internet and historical monitoring data, the smoking behavior images are divided into three types of images including cigarette images, hand-held cigarette images and mouth smoking images, the images are respectively input into corresponding identification networks, loss values output by the three identification networks are subjected to comprehensive back propagation training, network parameters are adjusted, and the trained cigarette identification network, hand-held cigarette identification network and mouth smoking identification network are obtained.
And further, inputting the target monitoring images of the time sequence into a preset smoking recognition deep neural model according to the time sequence, and performing feature extraction on any one target monitoring image through the smoking recognition deep neural model to obtain a feature image of the target monitoring image. In this embodiment, a feature image of the target monitoring image is extracted by using a histogram of oriented gradients feature extraction algorithm.
S30: and according to the characteristic image of the target monitoring image, calculating smoking characteristic similarity in the characteristic image, and taking the smoking characteristic similarity as smoking behavior probability.
In this embodiment, the smoking feature similarity refers to the inverse of the distance measure between the smoking feature vector and the center vector in the feature image. The smoking behavior probability refers to the possibility of smoking behavior occurring in the target monitoring image.
Specifically, according to corresponding feature images acquired by a cigarette identification network, a hand-held cigarette identification network and a mouth smoking identification network, feature similarity related to smoking in the corresponding feature images is respectively calculated, smoking feature similarity in the feature images is calculated according to different weights, and the smoking feature similarity is used as the smoking behavior probability of the target monitoring image.
S40: and if the smoking behavior probability is greater than the preset threshold probability, generating smoking alarm information, and sending the smoking alarm information to a monitoring end associated with smoking.
In this embodiment, the threshold probability refers to a judgment base value set for the smoking behavior probability; the smoking alarm information is information for reminding smokers of forbidding smoking on a target floor; the monitoring end refers to a monitoring camera.
Specifically, according to the smoking behavior probability output by the smoking recognition deep neural model to the target monitoring image, if the smoking behavior probability is larger than a preset threshold probability, it is indicated that smoking behavior is occurring in the current monitoring area, smoking warning information is generated, and the smoking warning information is sent to a monitoring camera terminal associated with smoking to remind smokers that smoking is prohibited on the target floor. In this embodiment, the monitoring camera with the voice function is used for reminding the smokers of forbidding smoking on the target floor in a voice broadcast mode.
In this embodiment, the feature image of the target monitoring image includes a first feature image, a second feature image and a third feature image, as shown in fig. 2, that is, in step S20, the target monitoring image is input into a preset smoking recognition deep neural model according to a time sequence, and the obtaining of the feature image of the target monitoring image specifically includes:
s21: and acquiring a first characteristic image of the target monitoring image through a cigarette identification network.
In this embodiment, the first feature image is a local feature image with cigarettes extracted from the target monitoring image by the cigarette identification network.
Specifically, the cigarette identification network obtains an original cigarette characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original cigarette characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a first characteristic image of the target monitoring image.
S22: and acquiring a second characteristic image of the target monitoring image through the hand-held cigarette identification network.
In this embodiment, the second feature image is a local feature image with hand-held smoke extracted from the target monitoring image by the hand-held smoke recognition network.
Specifically, the hand-held cigarette identification network obtains an original hand-held cigarette characteristic image of the target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original hand-held cigarette characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a second characteristic image of the target monitoring image.
S23: and acquiring a third characteristic image of the target monitoring image through the mouth smoking identification network.
In this embodiment, the third feature image is a partial feature image with mouth smoking extracted from the target monitoring image by the mouth smoking recognition network.
Specifically, the mouth smoking identification network obtains an original mouth smoking characteristic image of the target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original mouth smoking characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a third characteristic image of the target monitoring image.
As shown in fig. 3, in step 30, calculating the smoking feature similarity in the feature image according to the feature image of the target monitoring image, and taking the smoking feature similarity as the smoking behavior probability includes:
s31: the cigarette identification network outputs the similarity containing the cigarette characteristic information in the first characteristic image through the radial basis function neural network layer, and the similarity containing the cigarette characteristic information is used as the probability of the first smoking behavior.
In this embodiment, the cigarette feature information is that the first feature image has a feature vector indicating a cigarette box or a cigarette.
Specifically, the output layer of the cigarette identification network is a radial basis function neural network layer, the radial basis function neural network layer calculates the distance from the feature vector to the center vector in the first feature image by using an Euclidean distance measurement method, the inverse quantity of the distance is used as the similarity of the first feature image containing the cigarette feature information, and the similarity containing the cigarette feature information is used as the first smoking behavior probability.
S32: the hand-held cigarette identification network outputs the similarity containing the hand-held cigarette characteristic information in the second characteristic image through the radial basis function neural network layer, and the similarity containing the hand-held cigarette characteristic information is used as a second smoking behavior probability.
In this embodiment, the hand-held cigarette feature information is that the second feature image has a feature vector of the finger-held cigarette.
Specifically, the output layer of the hand-held cigarette identification network is a radial basis function neural network layer, the radial basis function neural network layer calculates the distance from the feature vector to the central vector in the second feature image by using an Euclidean distance measurement method, the inverse quantity of the distance is used as the similarity with the hand-held cigarette feature information in the second feature image, and the similarity with the hand-held cigarette feature information is used as the second smoking behavior probability.
S33: and the mouth smoking identification network outputs the similarity containing the mouth smoking characteristic information in the third characteristic image through the radial basis function neural network layer, and takes the similarity containing the mouth smoking characteristic information as a third smoking behavior probability.
In the present embodiment, the mouth smoking feature information is that the third feature image has a feature vector indicating mouth smoking.
Specifically, the output layer of the mouth smoking identification network is a radial basis function neural network layer, the radial basis function neural network layer calculates the distance from the feature vector to the center vector in the third feature image by using an euclidean distance metric method, the inverse quantity of the distance is used as the similarity with the mouth smoking feature information in the third feature image, and the similarity with the mouth smoking feature information is used as the third smoking behavior probability.
S34: and setting the first smoking behavior probability as a first weight, setting the second smoking behavior probability as a second weight, setting the third smoking behavior probability as a third weight, and multiplying the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability by probability values obtained by corresponding weights respectively to serve as the smoking behavior probabilities. In this embodiment, the first weight is a proportion of the first smoking behavior probability when the smoking behavior probability is calculated; the second weight is the proportion of the second smoking behavior probability when the smoking behavior probability is calculated; the third weight is the proportion of the second smoking behavior probability when the smoking behavior probability is calculated.
Specifically, the third weight is set to be greater than the second weight, the second weight is greater than the first weight, and probability values obtained by multiplying the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability by the corresponding weights are used as the smoking behavior probabilities.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
In one embodiment, an intelligent building monitoring system is provided, which corresponds to the intelligent building monitoring methods in the above embodiments one to one. As shown in fig. 4, the intelligent building monitoring system includes an obtaining module 401, a feature extraction module 402, a probability calculation module 403, and a judgment and alarm module 404. The functional modules are explained in detail as follows:
the obtaining module 401 is configured to obtain a target monitoring image of a target floor.
The feature extraction module 402 is configured to input the target monitoring image into a preset smoking recognition deep neural model according to a time sequence, and obtain a feature image of the target monitoring image.
And a probability calculating module 403, configured to calculate a smoking feature similarity in the feature image according to the feature image of the target monitoring image, and use the smoking feature similarity as a smoking behavior probability.
And a judgment and alarm module 404, configured to generate smoking alarm information if the smoking behavior probability is greater than a preset threshold probability, and send the smoking alarm information to a monitoring end associated with smoking.
Furthermore, the intelligent building monitoring system of the embodiment further comprises an image preprocessing module,
and the image preprocessing module is used for carrying out normalization processing and data enhancement processing on the target monitoring image.
Further, the obtaining module 401 includes:
and the image acquisition sub-module is used for acquiring a target monitoring video of a target floor, decomposing the target monitoring video into a time sequence of single-frame images through OpenCV software, and taking the single-frame images as the target monitoring images.
Further, the feature extraction module 402 includes:
the first characteristic extraction submodule is used for acquiring a first characteristic image of a target monitoring image through a cigarette identification network;
the second feature extraction submodule is used for acquiring a second feature image of the target monitoring image through the hand-held cigarette identification network;
and the third feature extraction sub-module is used for acquiring a third feature image of the target monitoring image through the mouth smoking identification network.
Further, the first feature extraction sub-module includes:
and the first feature extraction unit is used for acquiring an original cigarette feature image of the target monitoring image by the cigarette identification network through each channel convolution layer in the network, performing data dimension reduction on the original cigarette feature image through the network pooling layer, and taking the image corresponding to the data dimension reduction as a first feature image of the target monitoring image.
Further, the second feature extraction sub-module includes:
and the second feature extraction unit is used for acquiring the original hand-held cigarette feature image of the target monitoring image by the hand-held cigarette identification network through the convolution layer of each channel in the network, performing data dimension reduction on the original hand-held cigarette feature image through the network pooling layer, and taking the image corresponding to the data dimension reduction as the second feature image of the target monitoring image.
Further, the third feature extraction sub-module includes:
and the third feature extraction unit is used for acquiring the original mouth smoking feature image of the target monitoring image through each channel convolution layer in the network by the mouth smoking identification network, performing data dimension reduction on the original mouth smoking feature image through a network pooling layer, and taking the image corresponding to the data dimension reduction as the third feature image of the target monitoring image.
Further, the probability calculation module 403 includes:
the first probability calculation submodule is used for outputting the similarity containing the cigarette characteristic information in the first characteristic image through the radial basis function neural network layer by the cigarette identification network, and taking the similarity containing the cigarette characteristic information as the probability of the first smoking behavior;
the second probability calculation submodule is used for outputting the similarity containing the hand-held cigarette characteristic information in the second characteristic image through the radial basis function neural network layer by the hand-held cigarette identification network, and taking the similarity containing the hand-held cigarette characteristic information as a second smoking behavior probability;
and the third probability calculation submodule is used for outputting the similarity containing the mouth smoking characteristic information in the third characteristic image through the radial basis function neural network layer by the mouth smoking identification network, and taking the similarity containing the mouth smoking characteristic information as a third smoking behavior probability.
Further, the probability calculation module 403 further includes:
and the comprehensive probability calculation submodule is used for setting the first smoking behavior probability as a first weight, setting the second smoking behavior probability as a second weight, setting the third smoking behavior probability as a third weight, and multiplying the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability by the probability values obtained by the corresponding weights respectively to obtain the smoking behavior probability.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 5. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing target monitoring images, characteristic images, smoking recognition deep neural models and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to realize the intelligent building monitoring method, and the processor executes the computer program to realize the following steps:
s10: acquiring a target monitoring image of a target floor;
s20: inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
s30: according to the feature image of the target monitoring image, smoking feature similarity in the feature image is calculated, and the smoking feature similarity is used as smoking behavior probability;
s40: and if the smoking behavior probability is greater than the preset threshold probability, generating smoking alarm information, and sending the smoking alarm information to a monitoring end associated with smoking.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
s10: acquiring a target monitoring image of a target floor;
s20: inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
s30: according to the feature image of the target monitoring image, smoking feature similarity in the feature image is calculated, and the smoking feature similarity is used as smoking behavior probability;
s40: and if the smoking behavior probability is greater than the preset threshold probability, generating smoking alarm information, and sending the smoking alarm information to a monitoring end associated with smoking.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), 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 a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. An intelligent building monitoring method, characterized in that the method comprises:
acquiring a target monitoring image of a target floor;
inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
according to the feature image of the target monitoring image, smoking feature similarity in the feature image is calculated, and the smoking feature similarity is used as smoking behavior probability;
and if the smoking behavior probability is greater than the preset threshold probability, generating smoking warning information, and sending the smoking warning information to a monitoring terminal associated with smoking.
2. The intelligent building monitoring method according to claim 1, wherein the step of obtaining the target monitoring image of the target floor comprises:
the method comprises the steps of obtaining a target monitoring video of a target floor, decomposing the target monitoring video into a single-frame image of a time sequence through OpenCV software, and taking the single-frame image as the target monitoring image.
3. The building intelligent monitoring method according to claim 1, wherein before the step of inputting the target monitoring image into the preset smoking recognition deep neural model in a time sequence, the method further comprises:
and carrying out normalization processing and data enhancement processing on the target monitoring image.
4. The building intelligent monitoring method according to claim 1, wherein the preset smoking recognition deep neural model comprises a cigarette recognition network, a hand-held cigarette recognition network and a mouth smoking recognition network; the characteristic images comprise a first characteristic image, a second characteristic image and a third characteristic image; the step of inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image comprises the following steps:
acquiring a first characteristic image of a target monitoring image through the cigarette identification network;
acquiring a second characteristic image of a target monitoring image through the hand-held cigarette identification network;
and acquiring a third characteristic image of the target monitoring image through the mouth smoking identification network.
5. The intelligent building monitoring method according to claim 4, wherein the step of obtaining the first characteristic image of the target monitoring image through the cigarette identification network comprises:
the cigarette identification network obtains an original cigarette characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original cigarette characteristic image through a network pooling layer, and takes an image corresponding to the data dimension reduction as a first characteristic image of the target monitoring image.
6. The intelligent building monitoring method according to claim 4, wherein the step of acquiring the second characteristic image of the target monitoring image through the hand-held smoke recognition network comprises the following steps:
the hand-held cigarette identification network obtains an original hand-held cigarette characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original hand-held cigarette characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a second characteristic image of the target monitoring image.
7. The building intelligent monitoring method according to claim 4, wherein the step of obtaining the third feature image of the target monitoring image through the mouth smoking identification network comprises:
the mouth smoking identification network obtains an original mouth smoking characteristic image of a target monitoring image through each channel convolution layer in the network, performs data dimension reduction on the original mouth smoking characteristic image through a network pooling layer, and takes a corresponding image after the data dimension reduction as a third characteristic image of the target monitoring image.
8. The building intelligent monitoring method according to claim 4, wherein the step of calculating smoking feature similarity in the feature image according to the feature image of the target monitoring image and taking the smoking feature similarity as smoking behavior probability comprises the steps of:
the cigarette identification network outputs the similarity containing the cigarette characteristic information in the first characteristic image through a radial basis function neural network layer, and the similarity containing the cigarette characteristic information is used as the probability of the first smoking behavior;
the hand-held cigarette identification network outputs the similarity containing the hand-held cigarette characteristic information in the second characteristic image through the radial basis function neural network layer, and takes the similarity containing the hand-held cigarette characteristic information as a second smoking behavior probability;
and the mouth smoking identification network outputs the similarity containing the mouth smoking characteristic information in the third characteristic image through the radial basis function neural network layer, and takes the similarity containing the mouth smoking characteristic information as a third smoking behavior probability.
9. The building intelligent monitoring method according to claim 8, wherein the method comprises the steps of calculating smoking feature similarity in the feature image according to the feature image of the target monitoring image, and taking the smoking feature similarity as smoking behavior probability, and further comprising:
and setting the first smoking behavior probability as a first weight, setting the second smoking behavior probability as a second weight, setting the third smoking behavior probability as a third weight, and multiplying the first smoking behavior probability, the second smoking behavior probability and the third smoking behavior probability by probability values obtained by corresponding weights respectively to serve as the smoking behavior probabilities.
10. An intelligent building monitoring system, the system comprising:
the acquisition module (401) is used for acquiring a target monitoring image of a target floor;
the characteristic extraction module (402) is used for inputting the target monitoring image into a preset smoking recognition deep neural model according to a time sequence to obtain a characteristic image of the target monitoring image;
a probability calculation module (403) for calculating smoking feature similarity in the feature image according to the feature image of the target monitoring image, and taking the smoking feature similarity as smoking behavior probability;
and the judgment warning module (404) is used for generating smoking warning information if the smoking behavior probability is greater than a preset threshold probability, and sending the smoking warning information to a monitoring terminal associated with smoking.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010853100.4A CN112036279A (en) | 2020-08-22 | 2020-08-22 | Intelligent building monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010853100.4A CN112036279A (en) | 2020-08-22 | 2020-08-22 | Intelligent building monitoring method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112036279A true CN112036279A (en) | 2020-12-04 |
Family
ID=73580576
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010853100.4A Pending CN112036279A (en) | 2020-08-22 | 2020-08-22 | Intelligent building monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112036279A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113625604A (en) * | 2021-07-30 | 2021-11-09 | 季伟 | Intelligent building safety protection method and device |
| CN113971839A (en) * | 2021-12-23 | 2022-01-25 | 宏景科技股份有限公司 | Method, device, equipment and medium for detecting abnormal human body behaviors in images and videos |
| CN115440015A (en) * | 2022-08-25 | 2022-12-06 | 深圳泰豪信息技术有限公司 | Video analysis method and system capable of being intelligently and safely controlled |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107396095A (en) * | 2017-08-28 | 2017-11-24 | 方玉明 | One kind is without with reference to three-dimensional image quality evaluation method |
| CN108734125A (en) * | 2018-05-21 | 2018-11-02 | 杭州杰视科技有限公司 | A kind of cigarette smoking recognition methods of open space |
| CN109978907A (en) * | 2019-03-22 | 2019-07-05 | 南京邮电大学 | A kind of sitting posture of student detection method towards household scene |
-
2020
- 2020-08-22 CN CN202010853100.4A patent/CN112036279A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107396095A (en) * | 2017-08-28 | 2017-11-24 | 方玉明 | One kind is without with reference to three-dimensional image quality evaluation method |
| CN108734125A (en) * | 2018-05-21 | 2018-11-02 | 杭州杰视科技有限公司 | A kind of cigarette smoking recognition methods of open space |
| CN109978907A (en) * | 2019-03-22 | 2019-07-05 | 南京邮电大学 | A kind of sitting posture of student detection method towards household scene |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113625604A (en) * | 2021-07-30 | 2021-11-09 | 季伟 | Intelligent building safety protection method and device |
| CN113971839A (en) * | 2021-12-23 | 2022-01-25 | 宏景科技股份有限公司 | Method, device, equipment and medium for detecting abnormal human body behaviors in images and videos |
| CN115440015A (en) * | 2022-08-25 | 2022-12-06 | 深圳泰豪信息技术有限公司 | Video analysis method and system capable of being intelligently and safely controlled |
| CN115440015B (en) * | 2022-08-25 | 2023-08-11 | 深圳泰豪信息技术有限公司 | Video analysis method and system capable of being intelligently and safely controlled |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112036279A (en) | Intelligent building monitoring method and system | |
| US11403876B2 (en) | Image processing method and apparatus, facial recognition method and apparatus, and computer device | |
| CN110659397B (en) | Behavior detection method and device, electronic equipment and storage medium | |
| CN112418009B (en) | Image quality detection method, terminal equipment and storage medium | |
| CN110929648A (en) | Monitoring data processing method and device, computer equipment and storage medium | |
| CN111368758B (en) | Face ambiguity detection method, face ambiguity detection device, computer equipment and storage medium | |
| CN113239874B (en) | Behavior gesture detection method, device, equipment and medium based on video image | |
| CN111191532B (en) | Face recognition method and device based on construction area and computer equipment | |
| CN111814776B (en) | Image processing method, device, server and storage medium | |
| CN111476275A (en) | Target detection method based on picture recognition, server and storage medium | |
| CN116543334A (en) | Key frame extraction method, device, electronic equipment and storage medium | |
| CN110717357A (en) | Early warning method and device, electronic equipment and storage medium | |
| CN111899470A (en) | Human body falling detection method, device, equipment and storage medium | |
| CN111985340A (en) | Face recognition method and device based on neural network model and computer equipment | |
| CN111582179A (en) | Monitoring video analysis method and device, computer equipment and storage medium | |
| CN117234455B (en) | Intelligent control method and system for audio device based on environment perception | |
| CN114173094A (en) | Video monitoring method and device, computer equipment and storage medium | |
| CN112926616B (en) | Image matching method and device, electronic equipment and computer readable storage medium | |
| CN110895663A (en) | Two-wheel vehicle identification method and device, electronic equipment and monitoring system | |
| CN110837783A (en) | Automatic warning method, device and storage medium for suspicious people coming from outside in intelligent community | |
| CN116453056A (en) | Target detection model construction method and transformer substation foreign matter intrusion detection method | |
| CN114937302A (en) | Smoking identification method, device and equipment and computer readable storage medium | |
| CN111666973B (en) | Vehicle damage picture processing method and device, computer equipment and storage medium | |
| CN114241401A (en) | Abnormality determination method, apparatus, device, medium, and product | |
| CN114283360A (en) | Method, apparatus, device and medium for detecting electric vehicle in elevator |
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 |