CN116429769A - Fire-fighting cabinet glass breakage detection method, device, medium and equipment - Google Patents
Fire-fighting cabinet glass breakage detection method, device, medium and equipment Download PDFInfo
- Publication number
- CN116429769A CN116429769A CN202310315536.1A CN202310315536A CN116429769A CN 116429769 A CN116429769 A CN 116429769A CN 202310315536 A CN202310315536 A CN 202310315536A CN 116429769 A CN116429769 A CN 116429769A
- Authority
- CN
- China
- Prior art keywords
- fire
- font
- cabinet
- glass
- fighting
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 135
- 239000011521 glass Substances 0.000 title claims abstract description 130
- 238000013135 deep learning Methods 0.000 claims abstract description 47
- 238000000034 method Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims description 23
- 238000004590 computer program Methods 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 230000007547 defect Effects 0.000 abstract 1
- 238000007689 inspection Methods 0.000 description 6
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- 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
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8883—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges involving the calculation of gauges, generating models
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Chemical & Material Sciences (AREA)
- Quality & Reliability (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method based on deep learning the method for detecting the glass breakage of the fire-fighting cabinet, comprising the following steps: constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model; an image to be detected is obtained, inputting the image to be detected into the fire-fighting cabinet detection model, acquiring fire control cabinet information in the image to be detected; inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass; judging whether the fire-fighting cabinet glass is damaged according to the font information, and generating damage information. The invention effectively overcomes the defects that the image data of the glass breakage of the fire-fighting cabinet is difficult to collect the problem of high difficulty in data annotation, greatly improves the efficiency and accuracy of detecting the glass breakage of the fire-fighting cabinet.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a fire control cabinet glass breakage detection method, device, medium and equipment based on deep learning.
Background
In order to ensure that the fire-fighting equipment can timely play a role in protecting people's safety, related departments can regularly check the fire-fighting equipment of each unit to find whether the fire-fighting cabinet is abnormal or not. In general, the glass of the fire-fighting cabinet is damaged, and the equipment in the fire-fighting cabinet is abnormal.
In the prior art, for the detection of fire-fighting cabinet glass breakage, comprising: acquiring images by using a camera fixed in front of the fire-fighting cabinet, inputting the acquired images into a deep neural network for glass detection to obtain detection frame information, and judging whether the fire-fighting cabinet glass is damaged according to the detection frame information; or inputting the detection frame information into a deep learning network for image segmentation analysis to obtain glass region information, and judging whether the fire-fighting cabinet glass is damaged according to the glass region information; or inputting the detection frame information into a deep neural network to perform target edge detection analysis to obtain glass crack information, and judging whether the fire-fighting cabinet glass is damaged according to the glass crack information. However, the image data of the glass damage of the fire-fighting cabinet are difficult to collect, the data marking difficulty is high, and the glass damage recognition effect is inaccurate.
Disclosure of Invention
The embodiment of the invention provides a fire-fighting cabinet glass damage detection method, device, medium and equipment based on deep learning, which are used for solving the problems of difficult image data collection, difficult labeling and low damage identification accuracy in the fire-fighting cabinet glass damage detection in the prior art.
A fire-fighting cabinet glass breakage detection method based on deep learning, the method comprising:
constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
acquiring an image to be detected, inputting the image to be detected into the fire-fighting cabinet detection model, and acquiring fire-fighting cabinet information in the image to be detected;
inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass;
judging whether the fire-fighting cabinet glass is damaged according to the font information, and generating damage information.
Optionally, the constructing the fire cabinet detection model includes: the input picture resolution ratio of the deep learning YOLOv5 model is modified to 16:9, a step of performing the process;
and modifying the main network of the deep learning YOLOv5 model into a lightweight neural network SqueEzeNet, and eliminating the Focus structure in the deep learning YOLOv5 model to obtain a fire control cabinet detection original model.
And acquiring a fire-fighting cabinet training sample set, and performing multi-scale training on the fire-fighting cabinet detection original model by adopting the fire-fighting cabinet training sample set to obtain a fire-fighting cabinet detection model.
Optionally, the font information includes a font existing on fire-fighting cabinet glass and a confidence level of the font.
Optionally, the judging whether the fire-fighting cabinet glass is damaged according to the font information includes:
acquiring the due font number on the fire-fighting cabinet glass;
acquiring the number of the recognizable fonts on the fire-fighting cabinet glass according to the font information;
comparing the due font number with the recognizable font number;
if the number of the identifiable fonts is equal to the number of the due fonts, the fire-fighting cabinet glass is not damaged;
if the number of the identifiable fonts is smaller than the number of the due fonts, the fire-fighting cabinet glass is damaged.
Optionally, the acquiring the number of the recognizable fonts on the fire-fighting cabinet glass according to the font information includes:
initializing the recognizable font number to the due font number;
traversing each font existing on the fire-fighting cabinet glass, and comparing the confidence coefficient of the font with a preset confidence coefficient threshold;
and when the confidence coefficient of the fonts is smaller than a preset confidence coefficient threshold value, subtracting 1 from the number of the recognizable fonts.
Optionally, the due font number is obtained through a priori knowledge.
Optionally, the preset confidence threshold is 0.7 to 0.8.
A fire cabinet glass breakage detection method based on deep learning, the device comprising:
the construction module is used for constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
the fire control cabinet acquisition module is used for acquiring an image to be detected, inputting the image to be detected into the fire control cabinet detection model and acquiring fire control cabinet information in the image to be detected;
the font acquisition module is used for inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass;
and the damage judging module is used for judging whether the fire-fighting cabinet glass is damaged according to the font information and generating damage information.
A computer readable storage medium storing a computer program which when executed by a processor implements a fire department glass breakage detection method based on deep learning as described above.
A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing a fire-fighting cabinet glass breakage detection method based on deep learning as described above when executing the computer program.
According to the embodiment of the invention, the fire-fighting cabinet detection model and the fire-fighting cabinet font detection model are constructed; acquiring an image to be detected, inputting the image to be detected into the fire-fighting cabinet detection model, and acquiring fire-fighting cabinet information in the image to be detected; inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass; judging whether the fire-fighting cabinet glass is damaged according to the font information, generating damage information, greatly improving the efficiency and accuracy of fire-fighting cabinet glass damage detection, and effectively overcoming the problems that the fire-fighting cabinet glass damage image data are difficult to collect and the data marking difficulty is high.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a fire control cabinet glass breakage detection method based on deep learning according to an embodiment of the invention;
fig. 2 is a schematic structural diagram of a fire-fighting cabinet glass breakage detection device based on deep learning according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a computer device in accordance with an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The fire control cabinet glass breakage detection method based on deep learning, provided by the embodiment of the invention, is applied to a background server of a patrol robot, and a camera is arranged at the top of the patrol robot. The fire-fighting cabinet inspection route is preset, and the inspection robot is deployed to carry out patrol inspection along the fire-fighting cabinet inspection route. In the patrol process, when the patrol robot encounters the fire-fighting cabinet, the camera is automatically controlled to be aligned to the fire-fighting cabinet, surrounding image information is captured through the camera in real time, and the image information is transmitted to the background server in a 5G communication mode. The background server performs reasoning calculation and logic judgment of the deep neural network, acquires fire control cabinet information in an image to be detected, and acquires font information on fire control cabinet glass; judging whether the fire-fighting cabinet glass is damaged according to the font information, generating damage information, effectively improving the efficiency and accuracy of fire-fighting cabinet glass damage detection, and overcoming the problems that fire-fighting cabinet glass damage image data are difficult to collect and the difficulty of data marking is high.
The following describes the fire-fighting cabinet glass breakage detection method based on deep learning in detail, as shown in fig. 1, and the fire-fighting cabinet glass breakage detection method based on deep learning includes:
in step S101, a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model are constructed.
Here, the fire detection model is used to detect fire information in the input image, which is further provided to the fire font detection model. The fire control cabinet font detection model is used for detecting font information on fire control cabinet glass.
Alternatively, as a preferred example of the present invention, the constructing the fire cabinet detection model includes:
in step S201, the input picture resolution scale of the deep learning YOLOv5 model is modified to 16:9.
in step S202, the backbone network of the deep learning YOLOv5 model is modified into a lightweight neural network SqueezeNet, and the Focus structure in the deep learning YOLOv5 model is removed, so as to obtain a fire control cabinet detection original model.
In step S203, a fire-fighting cabinet training sample set is obtained, and Multiscale multi-scale training is performed on the fire-fighting cabinet detection original model by adopting the fire-fighting cabinet training sample set, so as to obtain a fire-fighting cabinet detection model.
Here, the present embodiment adopts the YOLOv5 model in deep learning to locate the fire cabinet in the image. By modifying the backbone network of the deep learning YOLOv5 model to a lightweight network, such as SqueezeNet, detection time can be effectively reduced. In model training, the embodiment adopts self-collection and labeling data as a fire cabinet training sample set. The method comprises the steps of obtaining a fire control cabinet detection original model by eliminating a Focus structure in a deep learning YOLOv5 model, and modifying the resolution ratio of pictures in a fire control cabinet training sample set to be 16: and 9, performing Multiscale multi-scale training on the fire control cabinet detection original model to obtain a fire control cabinet detection model, and effectively improving universality and monitoring capability of the model.
Alternatively, as a preferred example of the present invention, when constructing the fire-fighting cabinet font detection model, the present embodiment detects fonts on fire-fighting cabinet glass in an image output by the fire-fighting cabinet detection model using the YOLOv5 method in deep learning. In practice, fonts on fire-fighting cabinet glass include, but are not limited to, "fire," bolt, "" micro, "" station, "" police, "" 119. According to the embodiment, whether the fonts exist on the fire-fighting cabinet glass or not is obtained through the fire-fighting cabinet font detection model, and the confidence coefficient of each font is output. Optionally, during model training, the embodiment adopts self-collected and labeled data as a fire cabinet font training sample set, trains a neural network and acquires a weight file.
The trained fire-fighting cabinet detection model and the trained fire-fighting cabinet font detection model are used for carrying out fire-fighting cabinet detection and font detection on the image information shot by the inspection robot. The method further comprises the steps of:
in step S102, an image to be detected is acquired, the image to be detected is input to the fire fighting cabinet detection model, and fire fighting cabinet information in the image to be detected is acquired.
The fire control cabinet detection model marks the fire control cabinets in the image to be detected in a detection box mode, and fire control cabinet information is generated.
Optionally, in the stage of detecting the fire-fighting cabinet, i.e. the prediction stage, the present embodiment uses the YOLOv5 network model and the weight files trained during the training process. Firstly, preprocessing an image to be detected, such as: converting the picture to a fixed size 640 x 640; and then placing the image to be detected into the fire-fighting cabinet detection model, passing through a plurality of convolution layers and a pooling layer, and finally outputting a detection result through a full-connection layer. The detection results include, but are not limited to, a detection box of the fire fighting cabinet, and the confidence of the fire fighting cabinet. The fire-fighting cabinet confidence level indicates the probability that the information in the detection box is the fire-fighting cabinet. Optionally, the embodiment sets a fire-fighting cabinet confidence threshold, for example, above 0.5, and considers the content of the fire-fighting cabinet detection box to be a fire-fighting cabinet, and generates fire-fighting cabinet information.
In step S103, the fire fighting cabinet information is input to the fire fighting cabinet font detection model, and acquiring font information on the fire-fighting cabinet glass.
The font information comprises fonts existing on fire-fighting cabinet glass and the confidence of the fonts. After the fire control cabinet information is obtained in step S102, the fire control cabinet information is input into the fire control cabinet font detection model, and the fonts existing on the fire control cabinet glass and the confidence corresponding to the fonts are obtained through the fire control cabinet font detection model. One font corresponds to one confidence level.
Optionally, in the prediction stage, that is, detecting whether the fonts exist on the fire-fighting cabinet, using the fire-fighting cabinet font detection model and the weight file trained in the training process, inputting fire-fighting cabinet information into the fire-fighting cabinet font detection model, passing through a multi-layer convolution layer and a pooling layer, and finally passing through a full-connection layer prediction result. The prediction results comprise, but are not limited to, a font detection box and a font confidence on fire-fighting cabinet glass. The font confidence represents a probability that the content within the font detection box is a font. The closer the font detection box is to the actual font, the more the object is overlapped, and the higher the confidence value is. Therefore, the font confidence threshold is set, for example, to be 0.5, and when the font confidence is greater than 0.5, the content of the font detection box is considered to be a font.
In step S104, whether the fire-fighting cabinet glass is broken or not is judged according to the font information, and broken information is generated.
In the embodiment of the invention, whether the fire-fighting cabinet glass is missing or damaged is judged according to comprehensive analysis of the font information. Optionally, as a preferred embodiment of the present invention, the determining whether the fire-fighting cabinet glass is damaged according to the font information in step S104 includes:
in step S301, the number of fonts on the fire-fighting cabinet glass is obtained.
The due fonts are fonts contained in the fire-fighting cabinet glass in an intact state. For example, when the fire-fighting cabinet is shipped from a factory, the miniature fire-fighting station is displayed on glass, and the font of the fire-fighting cabinet is the miniature fire-fighting station. The due font number refers to the number of fonts contained in the fire-fighting cabinet glass in an intact state, and is obtained through priori knowledge.
In step S302, the number of recognizable fonts on the fire-fighting cabinet glass is obtained according to the font information.
The number of the recognizable fonts is the number of the recognizable fonts on the fire-fighting cabinet glass detected through deep learning. It should be understood that the recognizable fonts refer to complete fonts or incomplete fonts that reach a specified degree of recognition. The recognizable font number represents the font number actually contained on the fire-fighting cabinet glass. Here, the embodiment of the invention obtains the recognizable fonts and the number thereof on the fire-fighting cabinet glass based on the confidence degrees corresponding to the fonts. Optionally, as a preferred example of the present invention, the acquiring the number of identifiable fonts on the fire-fighting cabinet glass according to the font information includes:
in step S401, the recognizable font number is initialized to the due font number.
In step S402, each font existing on the fire-fighting cabinet glass is traversed, and the confidence level of the font is compared with a preset confidence level threshold.
In step S403, if the confidence level of the font is less than the preset confidence threshold, the number of recognizable fonts is reduced by 1.
Wherein the preset confidence threshold is 0.7 to 0.8, preferably 0.75. In this embodiment, a subtractive counting method is used to initialize the recognizable font number to the proper font number. And then traversing each font according to the font information obtained in the step S103, and comparing the confidence coefficient of the font with a preset confidence coefficient threshold value. When the confidence coefficient of the fonts is smaller than a preset confidence coefficient threshold value, the fonts are considered to be incomplete fonts, and the number of the recognizable fonts is reduced by 1; and when the confidence coefficient of the fonts is greater than or equal to a preset confidence coefficient threshold value, the fonts are considered to be recognizable fonts, and the number of the recognizable fonts is kept unchanged. After traversing each font in the font information, the number of the obtained identifiable fonts represents the number of the identifiable fonts on the fire-fighting cabinet glass in the image to be detected.
In step S303, the due font number and the recognizable font number are compared.
And comparing the due font number with the recognizable font number, and judging the size.
In step S304, if the number of identifiable fonts is equal to the number of font, the fire-fighting cabinet glass is not damaged.
In step S305, if the number of recognizable fonts is smaller than the number of due fonts, the fire-fighting cabinet glass is broken.
Comparing the due font number with the identifiable font number, and when the identifiable font number is equal to the due font number, indicating that the fonts on the fire-fighting cabinet glass in the image to be detected are completely identifiable, wherein the fire-fighting cabinet glass is not damaged and is complete; when the number of the identifiable fonts is smaller than the number of the due fonts, the fonts on the fire-fighting cabinet glass in the image to be detected are incomplete and are missing, and the fire-fighting cabinet glass is damaged.
To sum up, in the embodiment, the inspection robot is adopted to detect the damage of the fire-fighting cabinet, so that the cost of the whole system is effectively reduced, and the reusability is greatly improved. When the abnormal detection of the damage of the fire-fighting cabinet is carried out, as the glass of the fire-fighting cabinet is relatively close to the normal form, the missing and damaged state, the detection precision is lower when the target detection and the image segmentation are adopted, and the collection difficulty and the marking difficulty of the data of the damaged fire-fighting cabinet are high. In view of this, this embodiment combines font on the fire control cabinet glass and its confidence to carry out comprehensive judgement for the degree of difficulty that fire control cabinet glass was missed or was damaged to be detected reduces, has improved fire control cabinet damage detection's efficiency and rate of accuracy greatly, has overcome the problem that fire control cabinet damage image data is difficult to collect, the data mark degree of difficulty is big effectively.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.
In an embodiment, the invention further provides a fire-fighting cabinet glass breakage detection device based on deep learning, and the fire-fighting cabinet glass breakage detection device based on deep learning corresponds to the fire-fighting cabinet glass breakage detection method based on deep learning in the embodiment one by one. As shown in fig. 2, the fire-fighting cabinet glass breakage detection device based on deep learning comprises a construction module 21, a fire-fighting cabinet acquisition module 22, a font acquisition module 23 and a breakage judgment module 24. The functional modules are described in detail as follows:
a construction module 21, configured to construct a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
the fire control cabinet acquisition module 22 is configured to acquire an image to be detected, input the image to be detected into the fire control cabinet detection model, and acquire fire control cabinet information in the image to be detected;
the font obtaining module 23 is configured to input the fire fighting cabinet information to the fire fighting cabinet font detection model, and obtain font information on fire fighting cabinet glass;
and the breakage judgment module 24 is used for judging whether the fire-fighting cabinet glass is broken according to the font information, and generating breakage information.
Optionally, the fire control cabinet detection model is configured to detect fire control cabinet information in the input image, and the fire control cabinet information is further provided to the fire control cabinet font detection model. Optionally, in constructing the fire-fighting cabinet detection model, the construction module 21 is specifically configured to:
the input picture resolution ratio of the deep learning YOLOv5 model is modified to 16:9, modifying the backbone network into a lightweight neural network SqueezeNet, and eliminating a Focus structure in a deep learning YOLOv5 model to obtain a fire control cabinet detection original model;
and acquiring a fire control cabinet training sample set, and performing Multiscale multi-scale training on the fire control cabinet detection original model by adopting the fire control cabinet training sample set to obtain a fire control cabinet detection model.
Optionally, the fire-fighting cabinet font detection model is used for detecting font information on fire-fighting cabinet glass; in constructing the fire-fighting cabinet font detection model, the construction module 21 is specifically configured to:
using a YOLOv5 method in deep learning as a fire cabinet font detection original model;
and acquiring a fire-fighting cabinet font training sample set, and training the fire-fighting cabinet font detection original model by adopting the fire-fighting cabinet font training sample set to obtain a fire-fighting cabinet font detection model.
Optionally, the font information includes a font existing on fire-fighting cabinet glass and a confidence level of the font.
Here, after the image to be detected is acquired, the image to be detected is input into the fire-fighting cabinet detection model, and fire-fighting cabinet information marked in the image to be detected in a detection box mode is acquired. Then inputting the fire control cabinet information into the fire control cabinet font detection model; and acquiring fonts existing on the fire-fighting cabinet glass and confidence degrees corresponding to the fonts through the fire-fighting cabinet font detection model, wherein one font corresponds to one confidence degree.
Optionally, the breakage determination module 24 includes:
the first acquisition unit is used for acquiring the due font number on the fire-fighting cabinet glass;
the second acquisition unit is used for acquiring the number of the identifiable fonts on the fire-fighting cabinet glass according to the font information;
the comparison unit is used for comparing the due font number with the recognizable font number; if the number of the identifiable fonts is equal to the number of the due fonts, the fire-fighting cabinet glass is not damaged; if the number of the identifiable fonts is smaller than the number of the due fonts, the fire-fighting cabinet glass is damaged.
Optionally, the second acquisition unit includes:
an initialization subunit, configured to initialize the recognizable font number to a due font number;
the comparison subunit is used for traversing each font existing on the fire-fighting cabinet glass and comparing the confidence coefficient of the font with a preset confidence coefficient threshold value;
and the updating subunit is used for subtracting 1 from the number of the recognizable fonts if the confidence coefficient of the fonts is smaller than a preset confidence coefficient threshold value.
Optionally, the due font number is obtained through a priori knowledge.
Optionally, the preset confidence threshold is 0.7 to 0.8.
Here, in this embodiment, whether the fire-fighting cabinet glass is broken is determined by the font information. The method comprises the steps of obtaining the number of due fonts on fire-fighting cabinet glass, wherein the number of due fonts refers to the number of fonts contained in the fire-fighting cabinet glass in an intact state, and the number of due fonts is obtained through priori knowledge; initializing the recognizable font number to the due font number; traversing each font existing on the fire-fighting cabinet glass, and comparing the confidence coefficient of the font with a preset confidence coefficient threshold; if the confidence coefficient of the fonts is smaller than a preset confidence coefficient threshold value, subtracting 1 from the number of the recognizable fonts; comparing the number of the due fonts with the number of the identifiable fonts, judging the size, if the number of the identifiable fonts is equal to the number of the due fonts, the fire-fighting cabinet glass is not damaged, and if the number of the identifiable fonts is smaller than the number of the due fonts, the fire-fighting cabinet glass is damaged.
For specific limitations regarding the fire-fighting cabinet glass breakage detection device based on deep learning, reference may be made to the above limitation of the fire-fighting cabinet glass breakage method based on deep learning, and the detailed description thereof will be omitted. The modules in the fire control cabinet glass breakage device based on deep learning can be realized in whole or in part by software, hardware and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 3. 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 includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program when executed by the processor is used for realizing a fire control cabinet glass breakage detection method based on deep learning.
In one embodiment, a computer device is provided comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of when executing the computer program:
constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
acquiring an image to be detected, inputting the image to be detected into the fire-fighting cabinet detection model, and acquiring fire-fighting cabinet information in the image to be detected;
inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass;
judging whether the fire-fighting cabinet glass is damaged according to the font information, and generating damage information.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile 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), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the 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 scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (10)
1. The fire control cabinet glass breakage detection method based on deep learning is characterized by comprising the following steps of:
constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
acquiring an image to be detected, inputting the image to be detected into the fire-fighting cabinet detection model, and acquiring fire-fighting cabinet information in the image to be detected;
inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass;
judging whether the fire-fighting cabinet glass is damaged according to the font information, and generating damage information.
2. The fire cabinet glass breakage detection method based on deep learning of claim 1, wherein the constructing the fire cabinet detection model comprises:
the input picture resolution ratio of the deep learning YOLOv5 model is modified to 16:9, a step of performing the process;
and modifying the main network of the deep learning YOLOv5 model into a lightweight neural network SqueEzeNet, and eliminating the Focus structure in the deep learning YOLOv5 model to obtain a fire control cabinet detection original model.
And acquiring a fire-fighting cabinet training sample set, and performing multi-scale training on the fire-fighting cabinet detection original model by adopting the fire-fighting cabinet training sample set to obtain a fire-fighting cabinet detection model.
3. The deep learning-based fire department glass breakage detection method according to claim 1, wherein the font information includes a font existing on fire department glass and a confidence level of the font.
4. The deep learning-based fire cabinet glass breakage detection method according to claim 3, wherein the determining whether the fire cabinet glass is broken according to the font information comprises:
acquiring the due font number on the fire-fighting cabinet glass;
acquiring the number of the recognizable fonts on the fire-fighting cabinet glass according to the font information;
comparing the due font number with the recognizable font number;
if the number of the identifiable fonts is equal to the number of the due fonts, the fire-fighting cabinet glass is not damaged;
if the number of the identifiable fonts is smaller than the number of the due fonts, the fire-fighting cabinet glass is damaged.
5. The deep learning-based fire cabinet glass breakage detection method of claim 4, wherein the obtaining the number of identifiable fonts on the fire cabinet glass based on the font information comprises:
initializing the recognizable font number to the due font number;
traversing each font existing on the fire-fighting cabinet glass, and comparing the confidence coefficient of the font with a preset confidence coefficient threshold;
and if the confidence coefficient of the fonts is smaller than a preset confidence coefficient threshold value, subtracting 1 from the number of the recognizable fonts.
6. The fire cabinet glass breakage detection method based on deep learning of claim 5, wherein the due font number is obtained through a priori knowledge.
7. The deep learning-based fire cabinet glass breakage detection method of claim 5, wherein the preset confidence threshold is 0.7 to 0.8.
8. Fire control cabinet glass damage detection device based on degree of depth study, its characterized in that, the device includes:
the construction module is used for constructing a fire-fighting cabinet detection model and a fire-fighting cabinet font detection model;
the fire control cabinet acquisition module is used for acquiring an image to be detected, inputting the image to be detected into the fire control cabinet detection model and acquiring fire control cabinet information in the image to be detected;
the font acquisition module is used for inputting the fire control cabinet information into the fire control cabinet font detection model to acquire font information on fire control cabinet glass;
and the damage judging module is used for judging whether the fire-fighting cabinet glass is damaged according to the font information and generating damage information.
9. A computer-readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the deep learning-based fire department glass breakage detection method according to any one of claims 1 to 7.
10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the deep learning-based fire department glass breakage detection method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310315536.1A CN116429769A (en) | 2023-03-27 | 2023-03-27 | Fire-fighting cabinet glass breakage detection method, device, medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310315536.1A CN116429769A (en) | 2023-03-27 | 2023-03-27 | Fire-fighting cabinet glass breakage detection method, device, medium and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116429769A true CN116429769A (en) | 2023-07-14 |
Family
ID=87088328
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310315536.1A Pending CN116429769A (en) | 2023-03-27 | 2023-03-27 | Fire-fighting cabinet glass breakage detection method, device, medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116429769A (en) |
-
2023
- 2023-03-27 CN CN202310315536.1A patent/CN116429769A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111667011B (en) | Damage detection model training and vehicle damage detection method, device, equipment and medium | |
| CN112669316B (en) | Power production abnormality monitoring method, device, computer equipment and storage medium | |
| TWI716012B (en) | Sample labeling method, device, storage medium and computing equipment, damage category identification method and device | |
| CN110807491A (en) | License plate image definition model training method, definition detection method and device | |
| CN111310826B (en) | Method and device for detecting labeling abnormality of sample set and electronic equipment | |
| CN112613569B (en) | Image recognition method, training method and device for image classification model | |
| CN111325069B (en) | Production line data processing method and device, computer equipment and storage medium | |
| CN110955590A (en) | Interface detection method, image processing method, device, electronic equipment and storage medium | |
| CN110717449A (en) | Vehicle annual inspection personnel behavior detection method and device and computer equipment | |
| CN113343779B (en) | Environment abnormality detection method, device, computer equipment and storage medium | |
| CN116048124B (en) | Unmanned plane subway tunnel inspection method and device, computer equipment and storage medium | |
| CN111815576B (en) | Method, device, equipment and storage medium for detecting corrosion condition of metal part | |
| CN112001963A (en) | Fire fighting channel investigation method, system and computer equipment | |
| CN116964588A (en) | Target detection method, target detection model training method and device | |
| CN109684953B (en) | Method and device for pig tracking based on target detection and particle filter algorithm | |
| CN117372424B (en) | Defect detection method, device, equipment and storage medium | |
| CN114565780A (en) | Target identification method and device, electronic equipment and storage medium | |
| CN114972880A (en) | Label identification method and device, electronic equipment and storage medium | |
| CN111091104A (en) | Target object protection detection method, device, equipment and storage medium | |
| CN116863288A (en) | Target detection and alarm method, device and equipment based on deep learning | |
| CN116429769A (en) | Fire-fighting cabinet glass breakage detection method, device, medium and equipment | |
| CN116597431A (en) | Commodity outer package information detection method, commodity outer package information detection device, electronic equipment and storage medium | |
| CN116002480A (en) | Automatic detection method and system for accidental falling of passengers in elevator car | |
| CN115908977A (en) | Image data labeling method and device, electronic equipment and storage medium | |
| CN111225297A (en) | Broadband passive optical network port resource remediation method and system |
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 |