CN117253120A - Fire disaster identification method, device and storage medium - Google Patents

Fire disaster identification method, device and storage medium Download PDF

Info

Publication number
CN117253120A
CN117253120A CN202311247325.5A CN202311247325A CN117253120A CN 117253120 A CN117253120 A CN 117253120A CN 202311247325 A CN202311247325 A CN 202311247325A CN 117253120 A CN117253120 A CN 117253120A
Authority
CN
China
Prior art keywords
layer
convolution
image
feature
processing
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
Application number
CN202311247325.5A
Other languages
Chinese (zh)
Inventor
刘欣惠
窦瑞华
姜丁
马亮
燕永标
孟楠
王元杰
王立本
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Liantong Shandong Industry Internet Co ltd
China United Network Communications Group Co Ltd
Original Assignee
Liantong Shandong Industry Internet Co ltd
China United Network Communications Group Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Liantong Shandong Industry Internet Co ltd, China United Network Communications Group Co Ltd filed Critical Liantong Shandong Industry Internet Co ltd
Priority to CN202311247325.5A priority Critical patent/CN117253120A/en
Publication of CN117253120A publication Critical patent/CN117253120A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural 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
    • G06N3/04—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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
    • G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
    • G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Image Analysis (AREA)

Abstract

The application provides a fire disaster identification method, a fire disaster identification device and a fire disaster identification storage medium, and relates to the technical field of image processing. The method comprises the following steps: inputting an image to be identified into a convolutional neural network CNN model, and outputting a feature map set; inputting the feature image set into a coordinated attention CA model, and outputting attention feature data corresponding to the feature image set; and inputting the attention characteristic data set into the full-connection layer and the soft maximum Softmax layer, and outputting a fire disaster identification result which is used for indicating whether a fire disaster occurs in a position area corresponding to the image to be identified.

Description

Fire disaster identification method, device and storage medium
Technical Field
The present disclosure relates to the field of image processing technologies, and in particular, to a fire disaster identification method, device and storage medium.
Background
Fire has long been a significant threat to public and property security, and even life. With the rapid development of internet technology, fire identification technology is also advancing continuously to help people find fire more quickly.
In the prior art, the machine learning model carries out fire disaster identification through monitoring the real-time picture, and compared with the traditional identification method, the machine learning model has stronger identification capability and can judge the occurrence of fire disaster more accurately, thereby reducing the possibility of false alarm.
However, machine learning models also present some challenges in application. In the existing machine learning model, fire disaster identification can be completed based on a plurality of images or other large amount of data, the identification duration is long, the identification efficiency is low, and the occurrence of the fire disaster can not be identified timely. Therefore, how to improve the fire recognition efficiency is a highly desirable problem.
Disclosure of Invention
The application provides a fire disaster identification method, a fire disaster identification device and a storage medium, which are used for improving fire disaster identification efficiency.
In a first aspect, a fire identification method is provided, including: acquiring an image to be identified; inputting an image to be identified into a convolutional neural network (convolutional neural network, CNN) model, and outputting a feature map set; inputting the feature image set into a coordinated attention (coordinate attention, CA) model, and outputting attention feature data corresponding to the feature image set; and inputting the attention characteristic data set into a full connection layer and a soft maximum (Softmax) layer, and outputting a fire identification result which is used for indicating whether a fire occurs in a position area corresponding to the image to be identified.
The technical scheme provided by the application at least brings the following beneficial effects: after the images to be identified are acquired, the characteristic data in the images to be identified are rapidly extracted by using a convolutional neural network CNN model, so that the speed of processing the images to be identified is improved, the identification efficiency of fire is improved, the characteristic data in the images to be identified are further enhanced by coordinating an attention CA model, the accuracy of fire identification is guaranteed, the characteristic data are output in the form of attention characteristic data, and finally, a full-connection layer and a soft maximum Softmax layer judge whether fire occurs in a position area corresponding to the images to be identified according to the attention characteristic data, so that the identification efficiency of the fire is improved on the basis of guaranteeing the accuracy of fire identification, and the occurrence of the fire is timely identified.
As one possible implementation manner, the convolutional neural network CNN model includes a first feature processing layer, a second feature processing layer, a third feature processing layer and a fourth feature processing layer which are sequentially connected; the first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random deactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3 multiplied by 3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2 multiplied by 2; the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function, and a second random deactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the convolution kernels in the second convolution layer are all 3 multiplied by 3, the second pooling layer comprises 1 convolution kernel, and the convolution kernels in the second pooling layer are all 2 multiplied by 2; the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random deactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3 multiplied by 3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2 multiplied by 2; the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
As one possible implementation, the CA model includes a channel attention module and a spatial attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
As a possible implementation manner, the acquiring the image to be identified includes: collecting a video to be encoded; coding the video to be coded to obtain an initial image; the method comprises the steps of preprocessing an initial image to obtain an image to be identified, wherein the preprocessing comprises image enhancement processing, smoothing filtering processing and image sharpening processing.
As a possible implementation manner, the method further includes: and when the fire disaster identification result indicates that the fire disaster occurs in the position area corresponding to the image to be identified, sending out alarm information, wherein the alarm information is used for indicating the maintenance of the position area corresponding to the image to be identified.
In a second aspect, there is provided a fire identification device comprising: the acquisition module is used for acquiring the image to be identified; the processing module is used for inputting the image to be identified into the convolutional neural network CNN model and outputting a feature map set; the processing module is also used for inputting the feature image set into the coordination attention CA model and outputting attention feature data corresponding to the feature image set; the processing module is also used for inputting the attention characteristic data set into the full-connection layer and the soft maximum Softmax layer, outputting a fire disaster identification result, and indicating whether a fire disaster happens in a position area corresponding to the image to be identified or not.
As one possible implementation manner, the convolutional neural network CNN model includes a first feature processing layer, a second feature processing layer, a third feature processing layer and a fourth feature processing layer which are sequentially connected; the first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random deactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3 multiplied by 3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2 multiplied by 2; the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function, and a second random deactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the convolution kernels in the second convolution layer are all 3 multiplied by 3, the second pooling layer comprises 1 convolution kernel, and the convolution kernels in the second pooling layer are all 2 multiplied by 2; the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random deactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3 multiplied by 3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2 multiplied by 2; the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
As one possible implementation, the CA model includes a channel attention module and a spatial attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
As a possible implementation manner, the above-mentioned obtaining module is specifically configured to: collecting a video to be encoded; coding the video to be coded to obtain an initial image; the method comprises the steps of preprocessing an initial image to obtain an image to be identified, wherein the preprocessing comprises image enhancement processing, smoothing filtering processing and image sharpening processing.
As a possible implementation manner, the processing module is further configured to: and when the fire disaster identification result indicates that the fire disaster occurs in the position area corresponding to the image to be identified, sending out alarm information, wherein the alarm information is used for indicating the maintenance of the position area corresponding to the image to be identified.
In a third aspect, a fire identification device is provided, comprising a processor, which when executing a computer program implements the fire identification method according to the first aspect.
In a fourth aspect, a computer-readable storage medium is provided, the computer-readable storage medium comprising computer instructions; wherein the method of fire identification as described in the first aspect is implemented when the computer instructions are executed.
The advantageous effects described in the second aspect to the fourth aspect of the present invention may refer to the advantageous effect analysis of the first aspect, and are not described herein.
Drawings
The accompanying drawings are included to provide a further understanding of the technical aspects of the present application, and are incorporated in and constitute a part of this specification, illustrate the technical aspects of the present application and together with the examples of the present application, and not constitute a limitation of the technical aspects of the present application.
Fig. 1 is a schematic structural diagram of a fire disaster recognition system according to an embodiment of the present application;
fig. 2 is a flow chart of a fire disaster identification method according to an embodiment of the present application;
FIG. 3 is a flow chart of another fire disaster identification method according to an embodiment of the present application;
FIG. 4 is a workflow diagram of a deep flow toolkit provided in an embodiment of the present application;
fig. 5 is a schematic architecture diagram of a network model according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of a fire disaster recognition device according to an embodiment of the present application;
fig. 7 is a schematic structural diagram of another fire disaster recognition device according to an embodiment of the present application.
Detailed Description
The technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, but not all embodiments, and all other embodiments obtained by those skilled in the art without making creative efforts based on the embodiments in the present application are all within the scope of protection of the present application.
In the description of the present application, "/" means "or" unless otherwise indicated, for example, a/B may mean a or B. "and/or" herein is merely an association relationship describing an association object, and means that three relationships may exist, for example, a and/or B may mean: a exists alone, A and B exist together, and B exists alone. Furthermore, "at least one" means one or more, and "a plurality" means two or more. The terms "first," "second," and the like do not limit the number and order of execution, and the terms "first," "second," and the like do not necessarily differ. In this application, the terms "exemplary" or "such as" are used to mean serving as an example, instance, or illustration.
Any embodiment or design described herein as "exemplary" or "for example" should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "such as" is intended to present related concepts in a concrete fashion. In embodiments of the present application, "indication" may include both direct indication and indirect indication. For example, taking the first control information hereinafter as an example, the first control information may directly carry the information a itself or an index thereof, so as to achieve the purpose of directly indicating the information a. Alternatively, the first control information may also carry information B having an association relationship with information a, so as to achieve the purpose of indirectly indicating information a while indicating information B.
At present, related technologies for fire disaster recognition are mainly classified into the following four types:
1. the smoke identification technology is characterized in that smoke concentration change in a region is monitored through a smoke detector, and smoke characteristics are identified to judge whether fire disaster occurs. The technology has low input cost, but has low recognition accuracy, and the smoke detector is easy to be influenced by other gases to cause false alarm.
2. The optical recognition technology is to judge whether fire disaster occurs or not through the light change in the monitoring area of the infrared sensor and the light sensor. This technique is susceptible to light blockage resulting in false negatives.
3. The thermal imaging technology has the technical content that whether fire occurs in an area can be judged by monitoring the temperature change in the area and further acquiring a temperature image by using an infrared camera, and the position and the scale of a fire source can be determined according to the temperature image. However, this technique is high in input cost, and its recognition accuracy is relatively low due to the difference in heat generation property of different materials.
4. The machine learning model recognition technology, such as the background technology, is an advanced recognition technology in the related technology, and although the recognition accuracy can meet the normal requirement, the machine learning model recognition technology needs a large amount of data for judging whether a fire disaster occurs in an area, so that the machine learning model recognition technology has longer processing time and lower recognition efficiency, and can not timely recognize whether the fire disaster occurs in the area.
Based on the above, the present application provides a fire disaster identification method, which is characterized in that: after the images to be identified are obtained, the characteristic data in the images to be identified are rapidly extracted by using a convolutional neural network CNN model, so that the speed of processing the images to be identified is increased, the identification efficiency of fire is further improved, and the images are output in the form of a characteristic image set. The feature data in the image to be identified is further enhanced through the coordination attention CA model, the accuracy of fire disaster identification is guaranteed, the feature data are output in the form of attention feature data, and finally the full-connection layer and the soft maximum Softmax layer judge whether the fire disaster happens to the position area corresponding to the image to be identified according to the attention feature data, so that the fire disaster identification efficiency can be improved on the basis of guaranteeing the accuracy of fire disaster identification, and the occurrence of the fire disaster can be timely identified.
Embodiments of the present application will now be described with reference to the accompanying drawings.
Fig. 1 is a schematic structural diagram of a fire disaster recognition system according to an embodiment of the present application. The fire identification system comprises: a fire identification device 10 and an imaging device 20. The fire detection device 10 and the photographing device 20 may be connected by a wired or wireless connection.
The camera 20 may be disposed near the surveillance area. For example, taking a surveillance area as an example of a digital park, the photographing device 20 may be installed at a position where an image of the communication line and the area where the circuit is located can be photographed, such as on top of a building where the circuit and the communication line are located. The embodiment of the present application does not limit the specific installation manner and specific installation position of the photographing device 20.
The photographing device 20 may be used to photograph an image to be recognized of the surveillance area.
In some embodiments, camera 20 may employ a color camera to capture color images.
The color camera may be an RGB camera, for example. The RGB camera adopts an RGB color mode, and obtains various colors through the changes of three color channels of red (red, R), green (G), blue (B) and the superposition of the three color channels. Typically, an RGB camera gives three basic color components from three different cables, and three separate charge coupled device (charge coupled device, CCD) sensors are used to acquire the three color signals.
In some embodiments, camera 20 may employ a depth camera to capture depth images.
By way of example, the depth camera may be a time of flight (TOF) camera. The TOF camera adopts TOF technology, and the imaging principle of the TOF camera is as follows: the method comprises the steps of emitting modulated pulse infrared light according to a laser light source, reflecting the modulated pulse infrared light after encountering an object, receiving the light source reflected by the object by a light source detector, converting the distance between a TOF camera and a shot object by calculating the time difference or the phase difference between the emission and the reflection of the light source, and further obtaining the depth value of each point in a scene according to the distance between the TOF camera and the shot object.
The fire disaster recognition device 10 is configured to acquire an image to be recognized captured by the capturing device 20, and determine whether a fire disaster occurs in a location area corresponding to the image to be recognized based on the image to be recognized captured by the capturing device 20.
In some embodiments, the fire disaster identification device 10 may be an independent server, a server cluster or a distributed system formed by a plurality of servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks, and big data service networks.
In some embodiments, the fire identification device 10 may be a cell phone, tablet, desktop, laptop, handheld computer, notebook, ultra-mobile personal computer (UMPC), netbook, cell phone, personal digital assistant (personal digital assistant, PDA), augmented reality (augmented reality, AR)/Virtual Reality (VR) device, or the like.
In some embodiments, the fire identification device 10 may communicate with other terminal devices, such as with a terminal device of a worker, by wired or wireless means, to send an alarm message to the terminal device of the worker.
It should be appreciated that fig. 1 is an exemplary schematic diagram, and that the number of devices included in the fire identification system shown in fig. 1 is not limited, for example, the number of cameras is not limited. In addition to the equipment shown in fig. 1, the fire recognition system shown in fig. 1 may include other equipment, which is not limited thereto.
The execution main body of the fire disaster identification method provided by the embodiment of the application is a fire disaster identification device. Alternatively, the fire identification device may be the fire identification device 10 described above; alternatively, the fire recognition device may be a processor in the fire recognition device 10; still alternatively, the fire recognition device may be an Application (APP) for executing a fire recognition method installed in the fire recognition device 10; alternatively, the fire recognition device may be a functional module having an image processing function in the fire recognition device 10. This is not a limitation of the present examples.
Next, as shown in fig. 2, a flow chart of a fire disaster identification method according to an embodiment of the present application is provided, and the method includes the following steps:
s101, acquiring an image to be identified.
The image to be identified is an image obtained by photographing the surveillance area by the photographing device 20 in fig. 1. The supervision area is an area where it is necessary to supervise whether a fire occurs. Such as the area in the digital campus where the communication lines and circuits are located.
Alternatively, as shown in fig. 3, the obtaining the image to be identified may be specifically implemented as the following steps:
s1011, collecting the video to be encoded.
Optionally, in the present application, the capturing device 20 in fig. 1 captures a surveillance area to obtain a video to be encoded corresponding to the surveillance area.
Wherein. The video to be encoded is a video file after compression and encoding, and illustratively, the video to be encoded is in a video stream format and is composed of a series of time-ordered video frames, each video frame being a picture.
S1012, coding the video to be coded to acquire an initial image.
Optionally, the method and the device for encoding the video to be encoded can restore the video to be encoded into a plurality of supervision images corresponding to different times through encoding the video to be encoded by a preset deep stream tool kit (deep stream SDK). Each supervision image can be used as an initial image corresponding to the supervision area.
S1013, preprocessing the initial image to acquire an image to be identified.
Optionally, the application preprocesses the initial image through a preset deep stream tool kit (deep stream SDK) to obtain a processed initial image, namely an image to be identified. For example, the size of the image to be identified may be 64×64×3, or may be in another size format, which is not limited in this application.
Wherein the preprocessing includes an image enhancement process, a smoothing filter process, and an image sharpening process.
As shown in fig. 4, a workflow diagram of a deep-flow toolkit is provided in the present application. Firstly, coding the acquired video to be coded to obtain an initial image. The initial image is then preprocessed to obtain the image to be identified.
In this way, the image quality can be improved by preprocessing the initial image, so that the image is smoother and more continuous, and the image is more suitable for analysis and recognition, thereby improving the recognition accuracy and recognition efficiency of the fire.
S102, inputting the image to be identified into a convolutional neural network CNN model, and outputting a feature map set.
The convolutional neural network CNN model comprises a first feature processing layer, a second feature processing layer, a third feature processing layer and a fourth feature processing layer which are sequentially connected; the feature processing layer may also be referred to as a convolution block, such as a first convolution block, a second convolution block, and a third convolution block, for example.
The first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random deactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3 multiplied by 3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2 multiplied by 2;
the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function, and a second random deactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the convolution kernels in the second convolution layer are all 3 multiplied by 3, the second pooling layer comprises 1 convolution kernel, and the convolution kernels in the second pooling layer are all 2 multiplied by 2;
the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random deactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3 multiplied by 3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2 multiplied by 2;
the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
It should be noted that, the CNN model is used for extracting data features in the image to be identified and outputting the data features in the form of a feature map set. In order to meet the recognition accuracy, the CNN model used in the related art may include ten or even hundreds of feature processing layers, and even if only a small amount of images to be recognized are processed, a long processing time is required, which results in low recognition efficiency of fire.
The CNN model that this application provided only includes three characteristic treatment layer and a convolution layer, compares with the CNN model among the prior art, and the CNN model that this application provided is light-weight, is directed against the processing speed of waiting to discern the image, and the CNN model that this application provided compares the CNN model among the prior art and has great promotion, has improved the processing efficiency of waiting to discern the image, and then has improved the recognition efficiency of conflagration.
Illustratively, taking the CNN model provided by the present application as an example, the workflow of the CNN model is described. When the image to be identified enters the first feature processing layer, the first convolution layer carries out convolution processing on the image to be identified to obtain a first initial feature map set, wherein the first initial feature map set comprises 16 first initial feature maps. Next, the first activated ReLU function performs activation processing on the first initial feature map set, the first pooling layer performs average pooling processing on the activated first initial feature map set, and finally the first initial feature map set after the processing is input to the first random deactivated Dropout layer, so that the first feature map set is obtained, and the problem of excessive fitting is avoided.
When the first feature map set enters the second feature processing layer, the second convolution layer carries out convolution processing on the first feature map set to obtain a second initial feature map set, wherein the second initial feature map set comprises 32 second initial feature maps. And then, activating the second initial feature map set by the second activated ReLU function, carrying out average pooling on the activated second initial feature map set by the second pooling layer, and finally inputting the processed second initial feature map set into the second random inactivation Dropout layer to obtain a second feature map set.
When the second feature map set enters the third feature processing layer, the third convolution layer carries out convolution processing on the second feature map set to obtain a third initial feature map set, wherein the third initial feature map set comprises 64 third initial feature maps. And then, activating the third initial feature map set by the third activated ReLU function, carrying out average pooling on the activated third initial feature map set by the third pooling layer, and finally inputting the processed third initial feature map set into the third random inactivation Dropout layer to obtain a third feature map set.
When the second feature map set enters the fourth feature processing layer, the fourth convolution layer carries out convolution processing on the second feature map set to obtain a fourth initial feature map set, wherein the fourth initial feature map set comprises 128 fourth initial feature maps. Next, the fourth activated ReLU function performs activation processing on the fourth initial feature map set to obtain a feature map set.
It can be understood that the feature intensity in the feature map set is greater than the feature intensity in the third feature map set, the feature intensity in the third feature map set is greater than the feature intensity in the second feature map set, the feature intensity in the second feature map set is greater than the feature intensity in the first feature map set, and the greater the feature intensity of the feature map set is, the more suitable the analysis and recognition are, and the higher the recognition accuracy is.
S103, inputting the feature image set into a coordinated attention CA model, and outputting attention feature data corresponding to the feature image set.
The CA model comprises a channel attention module and a space attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
The CA model is used for extracting and enhancing the feature data in the feature map set, and the feature intensity of the obtained attention feature data is higher than the feature intensity in the feature map set after the channel attention processing and the space attention processing are carried out on the feature map set, so that the accuracy of the fire disaster recognition efficiency can be improved.
S104, inputting the attention characteristic data set into the full-connection layer and the soft maximum Softmax layer, and outputting a fire disaster identification result.
The full-connection layer can convert the received attention characteristic data into advanced characteristics, so that the soft maximum Softmax layer judges through the attention characteristic data, the probability of fire occurrence and the probability of no fire occurrence of a position area corresponding to the image to be identified are obtained, an event result with higher probability is input as a fire identification result, and the fire identification result is used for indicating whether fire occurs in the position area corresponding to the image to be identified.
Fig. 5 is a schematic diagram of a network model according to an embodiment of the present application. Taking the image size of the image to be identified as 64×64×3 as an example, after the image to be identified is input to the first feature processing layer, firstly, the image to be identified is subjected to convolution processing (Conv), then, the activation processing is performed by using an activation (ReLU) function, then, the average processing (Av-pool) is performed, and finally, the random inactivation processing (Dropout) is performed. The processing flows of the second processing layer to the fourth processing layer may refer to the processing flow of the image to be processed in S102, which is not described herein. After the fourth processing layer is processed, the feature image set obtained by processing is input into a coordinated attention CA model to obtain attention feature data, the attention feature data is input into a full connection layer (FC), and then is input into a soft maximum layer (Softmax), and finally a fire disaster recognition result is output.
In this way, after the image to be identified is obtained, the characteristic data in the image to be identified is quickly extracted by using the convolutional neural network CNN model, so that the speed of processing the image to be identified is increased, the identification efficiency of fire is further improved, and the characteristic data is output in the form of a characteristic image set. The feature data in the image to be identified is further enhanced through the coordination attention CA model, the accuracy rate of fire disaster identification is guaranteed, the feature data is output in the form of attention feature data, and finally the full-connection layer and the Softmax layer judge whether the fire disaster happens to the position area corresponding to the image to be identified according to the attention feature data, so that the fire disaster identification efficiency can be improved on the basis of guaranteeing the accuracy of fire disaster identification, and the occurrence of the fire disaster can be timely identified.
Based on the embodiment shown in fig. 2, after S104, the method further comprises: and sending out alarm information under the condition that the fire disaster identification result indicates that the fire disaster occurs in the position area corresponding to the image to be identified.
The alarm information is used for indicating the maintenance of the position area corresponding to the image to be identified.
For example, the alarm information may be in an information format, and may be directly sent to a terminal device of the staff member. The alarm information can also be in an audio format, and can be directly played on the audio equipment connected with the alarm information.
Therefore, under the condition that the fire disaster occurs in the position area corresponding to the image to be identified, workers can be timely reminded of maintaining the position area to reduce damage caused by the fire disaster.
The foregoing description of the solution provided in the embodiments of the present application has been mainly presented in terms of a method. To achieve the above functions, it includes corresponding hardware structures and/or software modules that perform the respective functions. Those of skill in the art will readily appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be implemented as hardware or combinations of hardware and computer software. Whether a function is implemented as hardware or computer software driven hardware depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
According to the embodiment of the application, the function modules of the fire disaster identification device can be divided according to the method example, for example, each function module can be divided corresponding to each function, and two or more functions can be integrated in one processing module. The integrated modules may be implemented in hardware or in software functional modules. Optionally, the division of the modules in the embodiments of the present application is schematic, which is merely a logic function division, and other division manners may be actually implemented.
Fig. 6 is a schematic structural diagram of a fire disaster identification device according to an embodiment of the present application, and as shown in fig. 6, the fire disaster identification device 60 includes: an acquisition module 601 and a processing module 602.
An acquisition module 601, configured to acquire an image to be identified;
the processing module 602 is configured to input an image to be identified to a convolutional neural network CNN model, and output a feature map set;
the processing module 602 is further configured to input the feature image set to the coordinated attention CA model, and output attention feature data corresponding to the feature image set;
the processing module 602 is further configured to input the attention feature data set to the full connection layer and the Softmax layer, and output a fire identification result, where the fire identification result is used to indicate whether a fire occurs in a location area corresponding to the image to be identified.
In some embodiments, the convolutional neural network CNN model includes a first feature processing layer, a second feature processing layer, a third feature processing layer, and a fourth feature processing layer connected in sequence; the first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random deactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3 multiplied by 3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2 multiplied by 2; the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function, and a second random deactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the convolution kernels in the second convolution layer are all 3 multiplied by 3, the second pooling layer comprises 1 convolution kernel, and the convolution kernels in the second pooling layer are all 2 multiplied by 2; the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random deactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3 multiplied by 3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2 multiplied by 2; the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
In some embodiments, the CA model includes a channel attention module and a spatial attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
In some embodiments, the acquiring module 601 is specifically configured to: collecting a video to be encoded; coding the video to be coded to obtain an initial image; the method comprises the steps of preprocessing an initial image to obtain an image to be identified, wherein the preprocessing comprises image enhancement processing, smoothing filtering processing and image sharpening processing.
In some embodiments, the processing module is further configured to: and when the fire disaster identification result indicates that the fire disaster occurs in the position area corresponding to the image to be identified, sending out alarm information, wherein the alarm information is used for indicating the maintenance of the position area corresponding to the image to be identified.
In the case of implementing the functions of the integrated modules in the form of hardware, the embodiment of the present application provides a fire identification device shown in fig. 7. As shown in fig. 7, the fire identification device 70 includes: processor 702, bus 704. Optionally, the fire identification device 70 may further include a memory 701; optionally, the fire identification device 70 may also include a communication interface 703.
The processor 702 may be any means for implementing or executing the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. The processor 702 may be a central processor, a general purpose processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. The processor 702 may also be a combination of computing functions, e.g., including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
A communication interface 703 for connecting with other devices via a communication network. The communication network may be an ethernet, a radio access network, a wireless local area network (wireless local area networks, WLAN), etc.
The memory 701 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (random access memory, RAM) or other type of dynamic storage device that can store information and instructions, or an electrically erasable programmable read-only memory (electrically erasable programmable read-only memory, EEPROM), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.
As a possible implementation, the memory 701 may exist separately from the processor 702, and the memory 701 may be connected to the processor 702 through the bus 704 for storing instructions or program codes. The processor 702, when calling and executing instructions or program code stored in the memory 701, can implement the fire identification method provided in the embodiments of the present application.
In another possible implementation, the memory 701 may also be integrated with the processor 702.
Bus 704, which may be an extended industry standard architecture (extended industry standard architecture, EISA) bus, or the like. The bus 704 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in fig. 7, but not only one bus or one type of bus.
The present application also provides a computer-readable storage medium including computer-executable instructions that, when executed on a computer, cause the computer to perform a method as provided in the above embodiments.
The present application also provides a computer program product directly loadable into a memory and including software code, which, when loaded and executed via a computer, is able to carry out the method provided by the above embodiments.
Those of skill in the art will appreciate that in one or more of the examples described above, the functions described herein may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, these functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
The foregoing is merely a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any changes or substitutions within the technical scope of the present disclosure should be covered in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (12)

1. A method of fire identification, the method comprising:
acquiring an image to be identified;
inputting the image to be identified into a convolutional neural network CNN model, and outputting a feature map set;
inputting the feature image set into a coordinated attention CA model, and outputting attention feature data corresponding to the feature image set;
and inputting the attention characteristic data set into a full-connection layer and a soft maximum Softmax layer, and outputting a fire disaster identification result, wherein the fire disaster identification result is used for indicating whether a fire disaster occurs in a position area corresponding to the image to be identified.
2. The method of claim 1, wherein the convolutional neural network CNN model comprises a first feature processing layer, a second feature processing layer, a third feature processing layer, and a fourth feature processing layer, connected in sequence;
the first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random inactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3×3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2×2;
the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function and a second random inactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the second convolution layer are 3×3, the second pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the second pooling layer are 2×2;
the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random inactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3×3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2×2;
the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
3. The method of claim 1, wherein the CA model comprises a channel attention module and a spatial attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
4. The method of claim 1, wherein the acquiring the image to be identified comprises:
collecting a video to be encoded;
coding the video to be coded to obtain an initial image;
and preprocessing the initial image to acquire the image to be identified, wherein the preprocessing comprises image enhancement processing, smoothing filtering processing and image sharpening processing.
5. The method according to any one of claims 1-4, further comprising:
and sending alarm information when the fire disaster identification result indicates that the position area corresponding to the image to be identified has a fire disaster, wherein the alarm information is used for indicating maintenance of the position area corresponding to the image to be identified.
6. A fire identification device, the device comprising:
the acquisition module is used for acquiring the image to be identified;
the processing module is used for inputting the image to be identified into a convolutional neural network CNN model and outputting a feature map set;
the processing module is also used for inputting the feature image set into a coordinated attention CA model and outputting attention feature data corresponding to the feature image set;
the processing module is further used for inputting the attention characteristic data set into the full connection layer and the soft maximum Softmax layer, and outputting a fire disaster identification result, wherein the fire disaster identification result is used for indicating whether a fire disaster occurs in a position area corresponding to the image to be identified.
7. The apparatus of claim 6, wherein the convolutional neural network CNN model comprises a first feature processing layer, a second feature processing layer, a third feature processing layer, and a fourth feature processing layer connected in sequence;
the first feature processing layer comprises a first convolution layer, a first pooling layer, a first activated ReLU function and a first random inactivated Dropout layer; the first convolution layer comprises 16 convolution kernels, the sizes of the convolution kernels in the first convolution layer are 3×3, the first pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the first pooling layer are 2×2;
the second feature processing layer comprises a second convolution layer, a second pooling layer, a second activated ReLU function and a second random inactivated Dropout layer; the second convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the second convolution layer are 3×3, the second pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the second pooling layer are 2×2;
the third feature processing layer comprises a third convolution layer, a third pooling layer, a third activated ReLU function and a third random inactivated Dropout layer; the third convolution layer comprises 32 convolution kernels, the sizes of the convolution kernels in the third convolution layer are 3×3, the third pooling layer comprises 1 convolution kernel, and the sizes of the convolution kernels in the third pooling layer are 2×2;
the fourth feature processing layer comprises a fourth convolution layer and a fourth activated ReLU function; the fourth convolution layer includes 128 convolution kernels, and the convolution kernels in the fourth convolution layer are each 6×6 in size.
8. The apparatus of claim 6, wherein the CA model comprises a channel attention module and a spatial attention module; the channel attention module is used for carrying out channel attention processing on the feature map set; the spatial attention module is used for performing spatial attention processing on the feature map set.
9. The apparatus of claim 6, wherein the obtaining module is specifically configured to:
collecting a video to be encoded;
coding the video to be coded to obtain an initial image;
and preprocessing the initial image to acquire the image to be identified, wherein the preprocessing comprises image enhancement processing, smoothing filtering processing and image sharpening processing.
10. The apparatus of any of claims 6-9, wherein the processing module is further configured to:
and sending alarm information when the fire disaster identification result indicates that the position area corresponding to the image to be identified has a fire disaster, wherein the alarm information is used for indicating maintenance of the position area corresponding to the image to be identified.
11. A fire identification device comprising a processor which when executing a computer program implements the fire identification method according to any one of claims 1 to 5.
12. A computer-readable storage medium, the computer-readable storage medium comprising computer instructions; wherein the computer instructions, when executed, implement the fire identification method of any one of claims 1 to 5.
CN202311247325.5A 2023-09-25 2023-09-25 Fire disaster identification method, device and storage medium Pending CN117253120A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311247325.5A CN117253120A (en) 2023-09-25 2023-09-25 Fire disaster identification method, device and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311247325.5A CN117253120A (en) 2023-09-25 2023-09-25 Fire disaster identification method, device and storage medium

Publications (1)

Publication Number Publication Date
CN117253120A true CN117253120A (en) 2023-12-19

Family

ID=89130805

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311247325.5A Pending CN117253120A (en) 2023-09-25 2023-09-25 Fire disaster identification method, device and storage medium

Country Status (1)

Country Link
CN (1) CN117253120A (en)

Similar Documents

Publication Publication Date Title
WO2020078229A1 (en) Target object identification method and apparatus, storage medium and electronic apparatus
CN109299703B (en) Method and device for carrying out statistics on mouse conditions and image acquisition equipment
US11195010B2 (en) Smoke detection system and method
CN113052029A (en) Abnormal behavior supervision method and device based on action recognition and storage medium
CN110580428A (en) image processing method, image processing device, computer-readable storage medium and electronic equipment
CN113887412B (en) Detection method, detection terminal, monitoring system and storage medium for pollution emission
CN113012383B (en) Fire detection alarm method, related system, related equipment and storage medium
JP2006174468A (en) Method and apparatus for image dispersion analysis
CN110796580B (en) Intelligent traffic system management method and related products
Fawzi et al. Embedded real-time video surveillance system based on multi-sensor and visual tracking
CN110728212B (en) Road well lid monitoring device and monitoring method based on computer vision
CN114627388B (en) Foreign matter detection equipment and foreign matter detection method for power transmission line
CN116246416A (en) Intelligent analysis early warning platform and method for security protection
CN107547839A (en) Remote control table based on graphical analysis
CN113158963B (en) Method and device for detecting high-altitude parabolic objects
CN105933676A (en) Remote control platform based on image analysis
CN112528825A (en) Station passenger recruitment service method based on image recognition
CN112016380A (en) Wild animal monitoring method and system
CN117253120A (en) Fire disaster identification method, device and storage medium
CN113283273A (en) Front obstacle real-time detection method and system based on vision technology
CN114095725B (en) Method and system for judging whether camera is abnormal
CN112435240B (en) Deep vision mobile phone detection system for workers to illegally use mobile phones
CN116597501A (en) Video analysis algorithm and edge device
CN113065416B (en) Leakage monitoring device integrated in transformer substation video monitoring device, leakage monitoring method and medium
CN112907454B (en) Method, device, computer equipment and storage medium for acquiring image

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