High-precision image recognition method and device in coal conveying gallery smoke and fire scene and medium

Abstract

The invention provides a high-precision image recognition method in a coal conveying gallery smoke and fire scene, which comprises the following steps of: 1, constructing equipment in the coal conveying gallery smoke and fire scene, and acquiring a real-time video data set of smoke and open fire in the coal conveying gallery smoke and fire scene; 2, pre-training the real-time video data set through a yolk v10 model, and dividing a smoke region and an open fire region; according to the method, the model library is used in advance for the coal conveying gallery scene of the thermal power plant, pre-training can be performed by directly utilizing the field video, and manual identification is performed after training, so that the processing period can be greatly shortened, and the defects in the prior art are overcome; according to the method, for real-time and big-data video files, only a small number of video files are selected for rapid recognition, compared with a traditional visual recognition technology which often needs a large number of manual marks to train a model, delay is reduced by 46%, and the method can be completed within several seconds; quick response is achieved, it is predicted that the fire occurrence rate can be reduced by 30% or above, and the response time can be completed within several seconds.

Description

Technical Field

[0001] The invention relates to a high-precision image recognition method, equipment and medium in a coal transportation corridor smoke and fire scene, and belongs to the field of image monitoring. Background Art

[0002] The bulk material conveying system of power plants and coal mines is a combination of a series of equipment and facilities used to transport, transfer and store bulk materials such as coal and fly ash within power plants and coal mines or between the two. The coal conveying corridor has a long span and the intensity of all-weather manual inspection is high. The dynamic perception ability of smoke and open flames generated by the conveying equipment is weak, and problems cannot be discovered in the first time. The coal slag and dust in the corridor are not easy to dissipate, which increases the risk of fire. As the national and local governments have gradually increased the safety requirements for coal transportation, relevant safety regulations and standards have been formulated, requiring the strengthening of fire risk management measures and the ability to achieve real-time monitoring.

[0003] However, due to the complex environment of coal mines, differences in technology from different manufacturers, as well as equipment size, equipment belt width, and space size, the accuracy of smoke and fire identification is reduced, and ordinary fire-fighting equipment has errors. The response time of common sprinkler systems is usually tens of seconds to several minutes. The coal conveying equipment generates fire and ignites the coal powder, and the handling efficiency is not high.

[0004] In view of this, the present invention provides a method, a computer device and a storage medium capable of detecting fireworks with high accuracy and rapidity. Summary of the invention

[0005] The present invention provides a method, a computer device and a storage medium capable of detecting fireworks with high accuracy and rapidity, which can effectively solve the above-mentioned problems.

[0006] The present invention is achieved in that:

[0007] A high-precision image recognition method, device and medium in a coal transportation corridor smoke and fire scene, comprising the following steps:

[0008] Step 1: Build multiple interrelated cameras, sprinkler equipment, and sound and light alarms in the coal transportation corridor fire scene to obtain a real-time video dataset of smoke and open flames in the coal transportation corridor fire scene;

[0009] Step 2: Pre-train the real-time video dataset using the Yo Lo V10 model to divide the smoke area and the open fire area;

[0010] Step 3, associating and configuring the camera at the same geographical location with the sprinkler head device;

[0011] Step 4, performing smoke and fire recognition on the open fire area, and dividing the smoke and fire recognition into different warning signals according to the smoke and fire warning threshold;

[0012] Step 5, calculate the bounding box values ​​of the smoke area and the open fire area in real time, and compare the bounding box values ​​with the smoke and fire warning threshold to obtain a smoke and fire identification warning signal to link the sound and light alarm or sprinkler head equipment.

[0013] Further, the step 1 specifically includes:

[0014] Step 11, according to the actual situation of the smoke and fire scene in the coal transportation corridor, deploy multiple visible light cameras, sound and light alarms and sprinkler head equipment; for example, in actual use, visible light cameras with a resolution of 1080p are arranged above the equipment operation area, and the distance between adjacent cameras is 20 meters; the actual situation of the smoke and fire scene in the coal transportation corridor described here includes but is not limited to the turning position and straight conveying position of the equipment to avoid visual blind spots;

[0015] Step 12, using multiple visible light cameras, sound and light alarms and sprinkler head equipment to obtain real-time video files.

[0016] Further, step 2 specifically includes:

[0017] Step 21, preprocessing the video files according to the set target number to obtain an image data set;

[0018] Step 22, configure the training parameters of the OpenCV pre-trained model, and set the missed recognition threshold, the precision threshold, and the missed recognition rate threshold;

[0019] Step 23, inputting the training data into the OpenCV pre-training model;

[0020] Step 24, during the start-up training process, smoke areas and open flame areas are identified according to the configured model training parameters, and the training process is monitored and verification data is input for verification;

[0021] Step 25, repeating step 24, by monitoring the training process, inputting verification data for verification, obtaining the number of correct recognitions, the number of false recognitions, the number of missed recognitions, and completing the initial marking;

[0022] Step 26, obtain the accuracy rate and missed recognition rate by calculation, and determine whether the model is put into use.

[0023] Furthermore, the step 21 specifically includes:

[0024] Step 211, a set target number of video files are read frame by frame through a video-to-picture conversion function, and converted to obtain a picture file having a picture sequence;

[0025] Step 212: divide the image file with the image sequence into training data and verification data according to a set ratio through a data set segmentation function, and mark them as label 1 and label 2 respectively.

[0026] Further, the step 3 specifically includes:

[0027] Step 31, numbering the cameras and sprinkler equipment in the smoke and fire scene of the coal transportation corridor so that each area where the identified camera is located is covered by a sprinkler equipment;

[0028] Step 32, manually input the serial numbers of the camera and the spray equipment into the system, or read the excel file through the read.excel reading function in the pandas function library.

[0029] Furthermore, the step 4 specifically includes: setting a fireworks warning threshold, and dividing fireworks recognition into three warning levels, namely a low warning level, a medium warning level, and a high warning level.

[0030] Further, the step 5 specifically includes:

[0031] Trigger the fireworks recognition warning signal and notify the on-site personnel immediately;

[0032] The if judgment function determines that the identified area is at the low warning level or the medium warning level, and calls the pymodbus library; the platform sends a request to the PLC, which can link the on-site sound and light alarm to notify the on-site personnel as soon as possible; among them, the pymodbus library is a standard library and an industrial communication protocol.

[0033] When the if judgment function determines that the alarm is at a high warning level, the platform automatically sends a switch signal through the modbus protocol to automatically start the sprinkler equipment associated with the area.

[0034] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the high-precision image recognition method in the smoke and fire scene of a coal transportation corridor is implemented.

[0035] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the high-precision image recognition method in the coal transportation corridor smoke and fire scene.

[0036] The beneficial effects of the present invention are as follows: the present invention uses a model library in advance for the coal transportation corridor scene of a thermal power plant, can directly use the on-site video for pre-training, and then perform manual marking after training, so that the processing cycle can be greatly shortened, making up for the shortcomings of the existing technology;

[0037] Compared with the earlier YOLO v9 and earlier versions of deep learning models, this method targets real-time and big data video files, and only selects a small number of video files for rapid recognition. Compared with traditional visual recognition technology, which often requires a large amount of manual labeling to train the model, the delay is reduced by 46% and can be completed within seconds; it achieves rapid response and is expected to reduce the incidence of fire by more than 30%, and the response time can be completed within seconds. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 It is an execution flow chart of a high-precision image recognition method in a coal transportation corridor smoke and fire scene provided by an embodiment of the present invention.

[0040] Figure 2 This is a flowchart for executing step 2 of a high-precision image recognition method in a coal transportation corridor smoke and fire scene provided by an embodiment of the present invention.

[0041] Figure 3 It is a schematic diagram of an electronic device provided by an embodiment of the present invention.

[0042] Figure 4 It is a schematic diagram of a computer-readable storage medium provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0044] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0045] Reference Figure 1 and Figure 2 As shown, this embodiment provides a specific implementation method for a high-precision image recognition method in a coal transportation corridor smoke and fire scene, comprising the following steps:

[0046] Step 1: Build multiple interrelated cameras, sprinkler equipment, and sound and light alarms in the coal transportation corridor fire scene to obtain a real-time video dataset of smoke and open flames in the coal transportation corridor fire scene;

[0047] Step 2: Pre-train the real-time video dataset using the Yo Lo V10 model to divide the smoke area and the open fire area;

[0048] Step 3, associating and configuring the camera at the same geographical location with the sprinkler head device;

[0049] Step 4, performing smoke and fire recognition on the open fire area, and dividing the smoke and fire recognition into different warning signals according to the smoke and fire warning threshold;

[0050] Step 5, calculate the bounding box values ​​of the smoke area and the open fire area in real time, and compare the bounding box values ​​with the smoke and fire warning threshold to obtain a smoke and fire identification warning signal to link the sound and light alarm or sprinkler head equipment.

[0051] In this embodiment, the step 1 specifically includes:

[0052] Step 11. According to the actual situation of the smoke and fire scene in the coal transportation corridor, deploy multiple visible light cameras, sound and light alarms and sprinkler head equipment. For example, in actual use, visible light cameras with a resolution of 1080p are arranged above the equipment operation area, and the distance between adjacent cameras is 20 meters. The actual situation of the smoke and fire scene in the coal transportation corridor described here includes but is not limited to the turning position and straight conveying position of the equipment to avoid visual blind spots. Among them, according to the different equipment targeted, the camera will be set at different heights to ensure that every angle of the coal transportation corridor can be captured and recorded. For example, for a low conveyor, the height of the camera only needs to be 3 meters, while for a higher elevator, a height of more than 8 meters needs to be set.

[0053] Step 12, using multiple visible light cameras, sound and light alarms and sprinkler head equipment to obtain real-time video files;

[0054] like Figure 2 As shown, in this embodiment, step 2 specifically includes:

[0055] Step 21, preprocessing the video files according to the set target number to obtain an image data set; specifically, it includes the following:

[0056] Step 211, a set target number of video files are read frame by frame through a video-to-picture conversion function, and converted to obtain a picture file having a picture sequence;

[0057] Step 212, dividing the image file with the image sequence into training data and verification data according to a set ratio by using a data set segmentation function, and marking them as label 1 and label 2 respectively;

[0058] For example, when the target number of video files is set to 1000, the 1000 video files are converted into video files with picture sequences through the video-to-picture function, and then divided into video files with a ratio of 8:2 through the dataset splitting function, where 80% of the video files with picture sequences are used for training and 20% of the video files with picture sequences are used for verification to ensure the adequacy of the training.

[0059] Among them, the video-to-picture function is a functional module whose purpose is to extract image frames from a given video resource according to specific rules and save these frames in the form of picture files, thereby realizing the conversion from video, a multimedia form containing continuous pictures, to discrete pictures.

[0060] Compared with traditional video recognition methods, which require manual calibration of thousands to tens of thousands of pictures, the time period required is too long and the efficiency cannot be guaranteed. The present invention adopts a video-to-picture function to read a set target number of video files instead of manually calibrating video files, which directly reduces the time of manual thinking and operation. The extracted graphic files are more accurate and regular, with small errors. When pre-training the model in the later stage, the source data is more objective, reducing the interference of subjective consciousness.

[0061] Step 22, configure the training parameters of the OpenCV pre-trained model, and set the missed recognition threshold, the precision threshold, and the missed recognition rate threshold;

[0062] Step 23, inputting the training data into the OpenCV pre-training model;

[0063] Step 24, during the start-up training process, smoke areas and open flame areas are identified according to the configured model training parameters, and the training process is monitored and verification data is input for verification;

[0064] Step 25, repeat step 24 to obtain the number of correct recognitions, the number of misrecognitions, and the number of missed recognitions, and complete the initial marking;

[0065] Step 26, by calculating the accuracy and missed recognition rate, determine whether the model is put into use, as follows:

[0066] If the accuracy reaches the precision threshold or is higher than the precision threshold and is less than the missed recognition rate, the model is put into use.

[0067] In this embodiment, the above process can be tested under multiple lighting scene conditions to verify the stability of the model. During the 24 hours of a day, the equipment receives different light intensities in the coal transportation corridor smoke and fire scene. For example, the sunlight intensity at noon is much greater than the light intensity at night. Excessive light intensity will cause incorrect identification of open fire areas.

[0068] In this embodiment, after step 24, during the model pre-training process, if an open fire area is identified, the present invention calls the requests library, sends an https request to the client, and reminds manual review. The labeling personnel correct it by themselves. The wrong labeling in the model pre-training, such as the wrong identification of open fire caused by strong light in the environment, is corrected by the personnel and the labeling is cancelled. The save function saves it to the database; if no wrong identification is detected during the training, it can be saved directly; and the evaluation phase is entered.

[0069] The specific steps for model pre-training are as follows:

[0070] In fact, 1,000 on-site pictures were collected for training, and the function calculation indicators are as follows:

[0071] TP: To correctly identify the number

[0072] FP: number of false positives

[0073] FN: Missed recognition number

[0074] Precision P = TP / (TP+FP+FN)

[0075] Missed recognition rate R = FN / (TP+FP+FN)

[0076] When the number of correct recognitions TP is 920, the number of false recognitions FP is 30, and the number of missed recognitions FN is 50, the precision P

[0077] 920 / 920+30+50=0.92

[0078] Then the missed recognition rate R is 50 / 920+30+50=0.05

[0079] If the accuracy reaches 0.85, i.e. 85%, or higher, and the missed recognition rate is less than 0.1, i.e. 10%, the model is put into use. The above process can be tested under multiple lighting scene conditions to verify the stability of the model.

[0080] The platform can set the missed recognition threshold (min_delta function) to 0.1. When the verification set accuracy is improved by less than 0.1, it proves that the missed recognition rate of model training is small. At this time, the callback function is automatically called to stop training.

[0081] In this embodiment, step 3 specifically includes:

[0082] Step 31, numbering the cameras and sprinkler equipment in the smoke and fire scene of the coal transportation corridor so that each area where the identified camera is located is covered by a sprinkler equipment;

[0083] In this embodiment, in actual application, it is necessary to collect the discharge information of the equipment in the smoke and fire scene of the coal transportation corridor and make it into an Excel file;

[0084] Camera number, such as A01

[0085] Camera position: conveyor belt head

[0086] Sprinkler No.: 001

[0087] Spraying device location: above the belt conveyor

[0088] Step 32, manually input the serial numbers of the camera and the spray equipment into the system, or read the Excel file through the read.excel reading function in the pandas function library, and automatically record them in the system.

[0089] In this embodiment, step 4 specifically includes:

[0090] Set the fire warning threshold and divide fire identification into three warning levels: low warning level, medium warning level, and high warning level;

[0091] Specifically:

[0092] Low warning level: smoke and open fire recognition area is less than 200 pixels;

[0093] Medium warning level: smoke and open fire recognition area is 200 to 500 pixels;

[0094] High warning level: smoke and open fire recognition area is greater than 500 pixels;

[0095] The specific method of dividing the level information means that since the Yo Lo V10 model in step 2 has calibrated the bounding box of the identified object during real-time monitoring, the reading function traverses the set target number of image and text files when actually reading the image, and the traversal function can obtain the model detection result. The traversal function described here is a function for accessing each element in the data structure; the main purpose is to systematically process each member of the set target number of image and text files to ensure that each image and text file is accessed, and usually in a certain order.

[0096] For example, in the actual operation process, the image taken by the camera is 1080p, and the recognition function can read the length and width of the bounding box. If the width w of the bounding box read is 20 pixels and the height H is 30 pixels, by W*H, the area of ​​the recognition area can be 600 pixels. This value is judged by the if judgment function, and the result is a high warning level;

[0097] In other embodiments, the fire and smoke warning threshold can also be set by the user, and the updated threshold is sent to the data bus platform by the http standard protocol, and the reading function is automatically called at runtime to ensure that the warning level can be flexibly configured and the threshold remains the latest value.

[0098] In this embodiment, step 5 specifically includes:

[0099] Trigger the fireworks recognition warning signal and notify the on-site personnel immediately;

[0100] If the judgment function determines that the identified area is at the low warning level or the medium warning level, the pymodbus library is called; the platform sends a request to the PLC, which can link the on-site sound and light alarm to notify the on-site personnel as soon as possible; among them, the pymodbus library is a standard library, which is an industrial communication protocol, and the PLC is a programmable logic controller.

[0101] When the if judgment function determines that the alarm is at a high warning level, the platform automatically sends a switch signal through the modbus protocol to automatically start the sprinkler equipment associated with the area.

[0102] In summary, the present invention uses a model library in advance for the coal transportation corridor scene of a thermal power plant, and can directly use the on-site video for pre-training, and then perform manual marking after training, which can greatly shorten the processing cycle and make up for the shortcomings of the existing technology;

[0103] Compared with the earlier YOLO v9 and earlier versions of deep learning models, this method targets real-time and big data video files, and only selects a small number of video files for rapid recognition. Compared with traditional visual recognition technology, which often requires a large amount of manual labeling to train the model, the delay is reduced by 46% and can be completed within seconds; it achieves rapid response and is expected to reduce the incidence of fire by more than 30%, and the response time can be completed within seconds.

[0104] like Figure 3 As shown, an embodiment of the present invention further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method when executing the program.

[0105] like Figure 4 As shown, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above method is implemented.

[0106] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0107] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0108] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0110] It should be noted that in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claim. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0111] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0112] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0113] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0114] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

Claims

1. A high-precision image recognition method for coal transportation corridor smoke and fire scenes, characterized in that: The following steps are included: Step 1: Build multiple interrelated cameras, sprinkler equipment, and sound and light alarms in the coal transportation corridor fire scene to obtain a real-time video dataset of smoke and open flames in the coal transportation corridor fire scene; Step 2: Pre-train the real-time video dataset using the YOLO v10 model to divide the smoke area and the open fire area; Step 3, associating and configuring the camera at the same geographical location with the sprinkler head device; Step 4, performing smoke and fire recognition on the open fire area, and dividing the smoke and fire recognition into different warning signals according to the smoke and fire warning threshold; Step 5, calculate the bounding box values ​​of the smoke area and the open fire area in real time, and compare the bounding box values ​​with the smoke and fire warning threshold to obtain a smoke and fire identification warning signal to link the sound and light alarm or sprinkler head equipment.

2. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as described in claim 1, characterized in that: The step 1 specifically includes: Step 11: deploy multiple visible light cameras, sound and light alarms, and sprinkler equipment according to the actual situation of the smoke and fire scene in the coal transportation corridor; Step 12, using multiple visible light cameras, sound and light alarms and sprinkler head equipment to obtain real-time video files.

3. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as described in claim 1, characterized in that: The step 2 specifically includes: Step 21, preprocessing the video files according to the set target number to obtain an image data set; Step 22, configure the training parameters of the OpenCV pre-trained model, and set the missed recognition threshold, the precision threshold, and the missed recognition rate threshold; Step 23, inputting the training data into the OpenCV pre-training model; Step 24, during the start-up training process, smoke areas and open flame areas are identified according to the configured model training parameters, and the training process is monitored and verification data is input for verification; Step 25, repeating step 24, by monitoring the training process, inputting verification data for verification, obtaining the number of correct recognitions, the number of false recognitions, the number of missed recognitions, and completing the initial marking; Step 26, obtain the accuracy rate and missed recognition rate by calculation, and determine whether the model is put into use.

4. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as described in claim 3, characterized in that: The step 21 specifically includes: Step 211, a set target number of video files are read frame by frame through a video-to-picture conversion function, and converted to obtain a picture file having a picture sequence; Step 212: divide the image file with the image sequence into training data and verification data according to a set ratio through a data set segmentation function, and mark them as label 1 and label 2 respectively.

5. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as described in claim 1, characterized in that: The step 3 specifically includes: Step 31, numbering the cameras and sprinkler equipment in the smoke and fire scene of the coal transportation corridor so that each area where the identified camera is located is covered by a sprinkler equipment; Step 32, manually input the serial numbers of the camera and the spray equipment into the system, or read the Excel file through the read.excel reading function in the pandas function library.

6. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as claimed in claim 1, characterized in that: The step 4 specifically includes: setting a fireworks warning threshold, and dividing fireworks recognition into three warning levels, namely a low warning level, a medium warning level, and a high warning level.

7. A high-precision image recognition method in a coal transportation corridor smoke and fire scene as claimed in claim 1, characterized in that: The step 5 specifically includes: Trigger the fireworks recognition warning signal and notify the on-site personnel immediately; The if judgment function determines that the identified area is at the low warning level or the medium warning level, and calls the pymodbus library; the platform sends a request to the PLC, which can link the on-site sound and light alarm to notify the on-site personnel as soon as possible; among them, the pymodbus library is a standard library and an industrial communication protocol. When the if judgment function determines that the alarm is at a high warning level, the platform automatically sends a switch signal through the modbus protocol to automatically start the sprinkler equipment associated with the area.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the high-precision image recognition method in the coal transportation corridor smoke and fire scene as described in any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the high-precision image recognition method in the coal transportation corridor smoke and fire scene as described in any one of claims 1 to 7 is implemented.

Images