Fire monitoring method and system and computer equipment

Abstract

The invention provides a fire monitoring method and system and computer equipment. The method comprises the following steps: acquiring a real-time image of a monitoring scene; calling a visual large model to analyze the real-time image, and generating fire description information of the real-time image through the visual large model under the condition that the visual large model detects that a smoke and fire target exists in the real-time image; and calling a large language model, and generating a rescue strategy according to the fire description information. According to the method, on one hand, the flame and smoke characteristics in the image can be identified by utilizing the capability of the visual large model, so that whether a fire occurs or not is accurately judged, and the false alarm rate can be effectively reduced; on the other hand, when it is judged that a fire occurs in the monitoring scene, fire description information is generated according to the real-time image. The fire description information serves as input of a large language model, and rescue suggestions are provided by means of natural language generation, inference analysis and knowledge association ability of the large language model. According to the method, the rescue action can be performed more quickly, and the rescue efficiency is improved.

Description

Technical Field

[0001] The present application relates to the fields of image processing technology and fire protection technology, and in particular to a fire monitoring method and system, and computer equipment. Background Art

[0002] Fire monitoring requires real-time and accurate acquisition of fire information, which is crucial to protecting the safety of life and property.

[0003] However, current fire monitoring and fire assessment methods face significant shortcomings. Monitoring accuracy is poor, as existing sensors are susceptible to interference from factors like fog and foreign objects, resulting in high false alarm rates and difficulty providing accurate fire data. Furthermore, fire assessments are severely delayed, requiring firefighters to arrive at the scene before assessing the fire's severity, source of the fire, and its spread, leading to delayed initial fire response. This is particularly true when dealing with unusual fire sources. Delayed information acquisition often prevents firefighters from preparing appropriate firefighting equipment in a timely manner, leading to missed opportunities for extinguishing the fire and posing significant challenges to disaster relief efforts. Summary of the Invention

[0004] In order to solve the existing technical problems, the present application provides a fire monitoring method and system, and computer equipment that can improve the accuracy of fire monitoring and the efficiency of fire rescue.

[0005] In a first aspect, a fire monitoring method is provided, the method comprising:

[0006] Obtain real-time images of the monitoring scene;

[0007] calling the visual big model to analyze the real-time image, and generating fire description information of the real-time image through the visual big model when the visual big model detects that there is a fire target in the real-time image;

[0008] A large language model is called to generate a rescue strategy based on the fire description information.

[0009] In a second aspect, a computer device is provided, comprising a processor and a memory connected to the processor, wherein the memory stores a computer program executable by the processor, and when the computer program is executed by the processor, the steps of the above-mentioned fire monitoring method are implemented.

[0010] On the third aspect, a fire monitoring system is provided, including: a fire dispatch server, a rescue terminal, and a camera and the above-mentioned computer equipment set up at the monitoring site, the camera is communicatively connected to the computer equipment, the computer equipment is communicatively connected to the fire dispatch server; the rescue terminal is communicatively connected to the fire dispatch server.

[0011] In a fourth aspect, a fire monitoring system is provided, comprising: a fire dispatch server, a rescue terminal, the above-mentioned computer equipment, and a camera and a main control terminal set up at the monitoring site, wherein the camera is communicatively connected to the main control terminal, the main control terminal at each monitoring site is communicatively connected to the computer equipment, and the computer equipment is communicatively connected to the fire dispatch server; the rescue terminal is communicatively connected to the computer equipment or to the fire dispatch server.

[0012] The fire monitoring method provided in the above embodiment uses a large visual model to analyze real-time images of the monitored scene to determine whether a fire has occurred in the monitored scene. If a fire has occurred, the large visual model generates fire description information for the real-time image. This method leverages the capabilities of the large visual model to identify flame and smoke features in the image, thereby accurately determining whether a fire has occurred and effectively reducing the false alarm rate. Furthermore, if a fire has occurred in the monitored scene, fire description information is generated based on the real-time image. By expressing fire elements, scene descriptions, and surrounding environment descriptions in text form, this method provides comprehensive and accurate real-time data support for fire monitoring, fire assessment, and subsequent rescue decision-making. Furthermore, this fire description information is used as input to a large language model, leveraging the model's natural language generation, reasoning analysis, and knowledge association capabilities to generate a rescue strategy. This method enables the use of the large visual model and large language model to generate a rescue strategy based on the real-time image as soon as a fire is detected (e.g., in the early stages of a fire). This allows disaster relief decisions to be made earlier in the fire's history, reducing the blindness of firefighting and rescue operations and improving the scientific nature of rescue efforts. This method enables rescue operations to be carried out more quickly, improves rescue efficiency, and minimizes the losses caused by fire.

[0013] The computer device and fire monitoring system provided in the above embodiments belong to the same concept as the corresponding fire monitoring method embodiments, and thus have the same technical effects as the corresponding fire monitoring method embodiments, and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 FIG. 1 is a schematic diagram of the system architecture of a fire monitoring system in one embodiment.

[0015] Figure 2 FIG. 1 is a schematic diagram of the system architecture of a fire monitoring system in another embodiment.

[0016] Figure 3 1 is a flow chart of steps of a fire monitoring method in one embodiment.

[0017] Figure 4 FIG. 4 is a flow chart of steps of a fire monitoring method in another embodiment. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0020] In the following description, the expression "some embodiments" is involved, which describes a subset of all possible embodiments. It should be noted that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict.

[0021] In the following description, the terms "first, second, and third" are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understandable that "first, second, and third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0022] Fire monitoring is a task that must ensure real-time and accuracy. Timely judgment of the specific situation of the fire is also a key and difficult test for front-line disaster relief personnel for subsequent disaster relief work.

[0023] To ensure real-time and accurate fire monitoring, current technology relies on sensor-based technologies. These sensors are deployed in fire-prone areas to collect fire-related information (such as temperature and smoke). However, these technologies suffer from low detection accuracy, a large installation area, and susceptibility to interference from other factors (such as fog and haze, obstructions, etc.), which can lead to false alarms. False alarms require manual verification and elimination, which fails to meet current technical requirements for fire monitoring.

[0024] The work of judging a fire after it breaks out relies heavily on the on-the-spot judgment of firefighters. Firefighters can only make judgments after the fire breaks out and arrive at the scene. By then, the fire has already spread, and they cannot determine the specific circumstances of the fire, the burning objects, or the location of the fire, which adds difficulty to the firefighters' on-the-spot judgment and disaster relief work. Secondly, fire information, especially special burning objects (such as flammable and explosive objects and objects that cannot be extinguished with conventional water sources), requires the preparation of relevant special fire-fighting equipment before the fire truck is dispatched. Failure to make relevant preparations in the first place will delay the relevant disaster relief process.

[0025] In view of the above problems, this application provides a fire monitoring system, such as Figure 1 As shown, a fire dispatch server 40, a rescue terminal 50, and a camera 10 and a main control device 20 set up at the monitoring site.

[0026] Specifically, each monitoring scene is equipped with a camera 10 and a main control device 20. The main control device 20 in each monitoring scene is connected to a fire dispatch server 40, which can communicate with multiple rescue terminals 50. The fire dispatch server 40 is used to implement fire dispatch. The rescue terminals 50 are used to receive dispatch instructions and rescue messages.

[0027] Camera 10 can be a visible light camera that captures visible light images of the monitored scene in real time. Camera 10 can also be a dual-light camera equipped with both visible light and infrared imaging systems, capable of capturing both visible light and infrared images of the same scene simultaneously or as needed. Camera 10 is connected to a host control device 20 and transmits real-time images of the monitored scene to the host control device 20.

[0028] The main control device 20 includes a processor and a memory connected to the processor. The memory stores a computer program that can be executed by the processor. When the computer program is executed by the processor, the fire monitoring method of the present application is implemented.

[0029] In each monitoring scenario, the master device 20 is also connected to at least one alarm device 30. When the master device 20 determines that a fire has occurred in the current scenario, it controls the alarm device 30 to sound an alarm. In some embodiments, the master device 20 is also connected to at least one fire extinguishing device 60 and controls the activation of the fire extinguishing device 60 to extinguish the fire.

[0030] The system may also include a display device and / or an alarm device connected to the fire dispatch server 40. The display device is configured to display at least one of the following: the fire location, real-time images, fire description information, rescue strategies, and rescue routes. The alarm device is configured to issue an alarm in response to fire or smoke detected in the real-time images.

[0031] This system architecture is suitable for a small, relatively independent fire monitoring system. Fire detection is performed by a master control device 20 at each monitoring site and communicates with a fire dispatch server 40. In this system architecture, the master control device 20 is located at the fire scene, close to the camera 10. It features fast response, enabling rapid acquisition and processing of image data, reducing data transmission time and bandwidth consumption. However, this also requires the master control device 20 to possess robust image processing and analysis capabilities, which places high demands on hardware configuration and increases equipment procurement costs.

[0032] In another embodiment, the structural diagram of a large fire monitoring system is as follows: Figure 2As shown, a camera 10 and a main control device 20 are provided at each monitoring site. The camera 10 is connected to the main control device 20 and sends real-time images of the monitoring scene to the main control device 20 .

[0033] Specifically, each monitoring scene is equipped with a camera 10 and a main control device 20. The main control device 20 of each monitoring scene is connected to a fire monitoring server 70 and sends real-time images to the fire monitoring server 70. The fire monitoring server 70 includes a processor and a memory connected to the processor. The memory stores a computer program executable by the processor. When the computer program is executed by the processor, it implements the fire monitoring method of the present application.

[0034] The fire monitoring server 70 is also in communication with the fire dispatch server 40. The rescue terminal 50 is also in communication with the fire monitoring server 70, or the rescue terminal 50 is in communication with the fire dispatch server 40. The fire dispatch server 40 is used to implement fire dispatch. The rescue terminal 50 is used to receive dispatch instructions and rescue messages.

[0035] The system may also include a display device and / or an alarm device connected to the fire dispatch server 40 and / or the fire monitoring server 70. The display device is configured to display at least one of the following: the fire location, real-time images, fire description information, rescue strategies, and rescue routes. The alarm device is configured to issue an alarm in response to detection of fire or smoke targets in the real-time images.

[0036] In this embodiment, real-time images are sent to the fire monitoring server 70 through the main control device 20, and the powerful computing power and comprehensive analysis advantages of the fire monitoring server 70 are utilized to improve the accuracy of fire judgment, support intelligent decision-making, and meet the application requirements of the fire monitoring system.

[0037] Based on the aforementioned fire monitoring system, a large visual model is invoked to analyze real-time images of the monitored scene to determine whether a fire has occurred. If a fire is determined to be present, the large visual model generates fire description information for the real-time image. This large visual model can leverage the capabilities of the large visual model to identify flame and smoke features in the image, accurately determining whether a fire has occurred and effectively reducing false alarm rates. Furthermore, if a fire is determined to be present in the monitored scene, a fire description is generated based on the real-time image. By expressing fire characteristics, scene descriptions, and surrounding environmental descriptions in text form, this system provides comprehensive and accurate real-time data support for fire monitoring, fire assessment, and subsequent rescue decision-making. Furthermore, this fire description information is used as input to a large language model, leveraging its natural language generation, reasoning analysis, and knowledge association capabilities to generate rescue strategies. This method enables the use of the large visual model and large language model to generate rescue strategies based on real-time images as soon as a fire is detected (e.g., in the early stages of a fire). This allows for disaster relief decisions to be made earlier in the fire's history, reducing the risk of blindness in firefighting and rescue operations and improving the scientific nature of rescue efforts. This method enables rescue operations to be carried out more quickly, improves rescue efficiency, and minimizes the losses caused by fire.

[0038] This application provides a fire monitoring method, such as Figure 3 As shown, it can be applied to Figure 1 The main control device 20 shown or Figure 2 The fire monitoring server 70 shown includes:

[0039] Step 302: Acquire a real-time image of the monitoring scene.

[0040] Specifically, a dual-light camera is used to capture the monitored scene, obtaining visible light and infrared images of the same scene. The dual-light camera can capture visible light and infrared video during the monitoring period, obtaining visible light and infrared images of the same scene. The dual-light camera can also capture visible light and infrared images at a preset frequency during the monitoring period, for example, capturing visible light and infrared images of the monitored scene every 1 second.

[0041] Step 204 : calling the visual big model to analyze the real-time image, and generating fire description information of the real-time image through the visual big model when the visual big model detects that there are fireworks in the real-time image.

[0042] A large visual model is an AI model for visual data such as images and videos. It is typically built on the Transformer framework, convolutional neural networks (CNNs), or a hybrid architecture, and can contain billions to hundreds of billions of parameters. It learns general visual feature representations by pre-training on massive amounts of multimodal data (such as images and text), and can quickly adapt to downstream tasks through fine-tuning or prompts. Commonly used large visual models include CLIP, SAM, and Stable Diffusion.

[0043] The visual big model of this application is based on the existing visual big model. By preparing training sample images that identify fireworks (including flames and smoke) targets and fire description information, the existing big model is fine-tuned to enable the visual big model to achieve the tasks of fire identification and fire description.

[0044] Compared with traditional fire monitoring methods such as traditional fire detectors, the visual big model can identify the flame and smoke characteristics in the image, thereby accurately determining whether a fire has occurred, which can effectively reduce the false alarm rate.

[0045] In this embodiment, Figure 1 In the system architecture shown, the main control device can be deployed with a large visual model. Figure 2 In the system architecture shown, the fire monitoring server can be deployed with a large visual model. Computer equipment ( Figure 1 The master device or Figure 2 The computer device) analyzes the fireworks targets in the real-time image by calling the visual big model to identify whether a fire has occurred in the monitoring scene through the real-time image, and generates fire description information of the real-time image when it is determined that a fire has occurred in the monitoring scene.

[0046] Fire description information utilizes advanced visual large-scale modeling technology to conduct deep intelligent understanding and analysis of real-time images of fire monitoring scenes, accurately converting key fire elements within the images into detailed text descriptions. This fire description includes descriptions of fire elements, the specific scene where the fire occurred, and the surrounding environment. Fire element descriptions include the nature, state, and quantity of burning objects, the presence of open flames, and significant amounts of smoke. The specific scene of the fire includes key information such as whether it was indoors or outdoors, and whether it involved crowded areas. The surrounding environment description includes information such as the presence of remaining combustible materials near the fire scene.

[0047] By expressing fire conditions, scene descriptions, and surrounding environment descriptions in text form, it provides comprehensive and accurate real-time data support for fire monitoring, fire assessment, and subsequent rescue decisions.

[0048] Step 206: Call the large language model to generate a rescue strategy based on the fire description information.

[0049] Specifically, the strategy generation prompt words and the fire description information are input into the large language model. The strategy generation prompt words are used to instruct the large language model to generate a rescue strategy according to the fire description information.

[0050] A large language model refers to a natural language processing (NLP) model based on deep learning. It is trained with massive text data and uses architectures such as Transformer to learn the statistical laws and semantic relationships of language. It can complete complex tasks such as text generation, question answering, reasoning, and translation.

[0051] In this embodiment, to enable the large language model to generate rescue strategies based on fire descriptions, the existing large language model can be fine-tuned using training samples. The training samples are fire descriptions labeled with rescue strategies, so that the fine-tuned large language model can generate rescue strategies based on the fire descriptions.

[0052] Specifically, the fire description information includes the description of fire elements, the specific scene of the fire, and the surrounding environment. Based on the fire description information, the large language model can provide rescue strategies in terms of fire-fighting operations, personnel evacuation, risk assessment and fire prediction.

[0053] For example, in terms of firefighting operations, recommendations can be made on the type of firefighting equipment to be used based on the type of burning object, and the number of police officers to be dispatched can be recommended based on the severity of the fire. In terms of risk assessment and fire prediction, predictions can be made on fire development, secondary disaster warnings, and rescue priorities based on wind direction, the type of combustible material, and the types of objects surrounding the combustible material.

[0054] This embodiment overcomes the drawback of traditional methods that require rescue personnel to be physically present at the scene to assess the fire situation. When a large visual model is used to determine a fire has occurred in the monitored scene, a large language model is then used to generate a rescue strategy using the fire description information from real-time images. This allows for pre-planned rescue strategies covering firefighting actions, evacuation, risk assessment, and fire spread prediction, reducing the blindness of firefighting and rescue operations and improving the scientific nature of rescue efforts. Rescuers can prepare accordingly based on the rescue strategy before dispatch, improving rescue efficiency.

[0055] The fire monitoring method described above uses a large visual model to analyze real-time images of the monitored scene to determine whether a fire has occurred. If a fire has occurred, the large visual model generates fire description information for the real-time image. This method leverages the capabilities of the large visual model to identify flame and smoke characteristics in the image, accurately determining whether a fire has occurred and effectively reducing false alarm rates. Furthermore, if a fire has occurred in the monitored scene, fire description information is generated based on the real-time image. By expressing fire elements, scene descriptions, and surrounding environment descriptions in text form, this method provides comprehensive and accurate real-time data support for fire monitoring, fire assessment, and subsequent rescue decision-making. Furthermore, this fire description information is used as input to a large language model, leveraging its natural language generation, reasoning analysis, and knowledge association capabilities to generate rescue strategies. This method enables the use of the large visual model and large language model to generate rescue strategies based on real-time images as soon as a fire is detected (e.g., in the early stages of a fire). This allows disaster relief decisions to be made earlier in the fire's history, reducing the blindness of firefighting and rescue operations and improving the scientific nature of rescue efforts. This method enables rescue operations to be carried out more quickly, improves rescue efficiency, and minimizes the losses caused by fire.

[0056] In one embodiment, if Figure 4 As shown, the fire monitoring method may include the following steps:

[0057] Step 402: Acquire a real-time image of the monitoring scene.

[0058] Step 404 : calling the visual big model to analyze the real-time image, and generating fire description information of the real-time image through the visual big model when the visual big model detects that there are fireworks in the real-time image.

[0059] The implementation process of this step is similar to step 304 and will not be repeated here.

[0060] Step 406: Determine the location of the fire based on the camera's installation location information and the relative relationship between the camera and the center of the fire.

[0061] The fire location refers to the latitude and longitude of the fire. This information allows the precise location of the fire, making it particularly useful in remote areas or those without clear addresses, such as mountainous areas and forests. Fire location information also provides information about the surrounding environment and population distribution, such as building types and the number of permanent residents. This helps understand the potential scope of the fire and surrounding hazards. Fire location information also provides information about surrounding facilities, including transportation, infrastructure, and firefighting facilities. This helps inform rescue routes, protective measures, and plans based on this information.

[0062] In step 408, the strategy generation prompt word, the fire location and the fire description information are input into the large language model. The strategy generation prompt word instructs the large language model to generate a rescue strategy according to the fire location and the fire description information.

[0063] The large language model learns from massive amounts of data and leverages its natural language generation, reasoning analysis, and knowledge association capabilities to provide rescue recommendations based on the location and description of the fire. For example, based on the location of the fire, the large language model can accurately locate the fire site, obtain information about the surrounding environment, population distribution, and surrounding facilities, and combine it with the fire description (including descriptions of fire elements, the specific scene of the fire, and the surrounding environment) to formulate a rescue strategy. Rescue strategies may include the following:

[0064] 1. Firefighting strategy. This includes recommending the appropriate type of fire extinguishing agent and number of personnel based on surrounding facility information and flammability characteristics, and developing firefighting tactics based on the topography and direction of fire spread.

[0065] 2. Evacuation strategy. This involves planning safe and rapid evacuation routes based on surrounding road conditions, traffic flow, and the terrain of the fire site. Evacuation priorities are also determined based on population distribution and risk levels.

[0066] 3. Rescue deployment strategy. This includes rationally deploying firefighting forces based on the scale of the fire, its spread, and the location and capabilities of surrounding fire stations. It also includes rationally deploying auxiliary rescue forces, taking into account other issues that may arise from the fire, such as medical care, power repair, and communications support.

[0067] 4. Secondary disaster prevention strategies. This involves assessing the risk of secondary disasters potentially caused by fire, such as water pollution, air pollution, and landslides, based on information about the surrounding environment. These strategies then develop corresponding preventative measures, as well as providing detailed safety precautions for rescue workers and affected residents at the fire scene and surrounding environment.

[0068] In this embodiment, at the early stage of a fire, a large language model can be used to generate a comprehensive rescue strategy based on the fire location and fire description information, thereby improving fire extinguishing efficiency, enhancing personnel safety, and improving the scientific nature of rescue decisions.

[0069] In one embodiment, a visual big model is called to analyze a real-time image, and when the visual big model detects the presence of fireworks and fire targets in the real-time image, the visual big model generates fire description information for the real-time image, including: inputting fire identification prompt words and the real-time image into the visual big model, the fire identification prompt words instructing the visual big model to detect fireworks and fire targets in the real-time image; when it is determined that the real-time image contains fireworks and fire targets, the fire description prompt words and the real-time image are input into the visual big model, the fire description prompt words instructing the visual big model to generate fire description information for the real-time image.

[0070] A prompt is a piece of guiding text used in large models to trigger the model to generate content in a specific direction or perform a special task.

[0071] In this embodiment, the large visual model is called to perform the fire detection task and the image description generation task.

[0072] The large-scale visual model detects fireworks in real-time images to determine whether a fire has occurred in the monitored scene. Preset fire identification prompts instruct the large-scale visual model to detect fireworks in the real-time image to verify whether a fire has occurred in the monitored scene. For example, the fire identification prompt could be: Please check for flames or smoke in the image.

[0073] The image description generation task involves using algorithms to analyze image content and generate corresponding textual representations. Pre-set description generation prompts instruct the visual model to analyze the real-time image and generate a fire description. For example, a description generation prompt might be: "Please describe the specific fire situation in the image, including the fire scene (indoor or outdoor), the location of the fire source, the burning object, the severity of the fire, whether it is in a crowded area, the presence of open flames and heavy smoke, and the presence of other combustible materials nearby."

[0074] In this embodiment, the visual big model is called twice to perform the fire detection task and the image description generation task respectively, and only when the fire detection task detects a fire is the visual big model called again to perform the image description generation task. This approach has many advantages.

[0075] On the one hand, fire detection is a relatively simple binary classification task. Large visual models consume few resources and are fast, enabling them to quickly identify images posing fire risks. On the other hand, image description generation is more complex, requiring the model to provide a detailed description of the fire in the image, including details such as fire size, flame color, and burning location. This undoubtedly consumes a significant amount of large visual model resources.

[0076] By adopting the above-mentioned step-by-step calling method, for real-time images where no fireworks targets are detected, there is no need to call the large visual model to perform the image description generation task, thereby avoiding consuming too many resources on a large number of irrelevant images, greatly saving the resource consumption of the large visual model, improving the efficiency of the entire process and the accuracy of resource utilization, and allowing model resources to focus more on the fire images that really need to be processed, ensuring the timeliness and accuracy of the fire description.

[0077] Traditional methods for locating fires require densely placed sensors, using the locations of alarm sensors to pinpoint the fire's location. This approach requires dense sensor placement and is costly. Traditional visual recognition methods, however, are unable to accurately pinpoint the fire's location.

[0078] In this embodiment, the GIS positioning system and camera-related information are utilized to accurately locate the location of the fire.

[0079] like Figure 4 As shown, step 406 determines the location of the fire based on the camera's installation location information and the relative relationship between the camera and the center of the fire, including:

[0080] Step 4061, control the camera to adjust and aim at the center of the fire, and obtain the pitch angle and azimuth angle of the camera when it is aimed at the center of the fire, as well as the latitude, longitude and altitude of the camera.

[0081] The fire center is the center of the fireworks target detected by the large visual model. By controlling the camera to adjust its focus on the fire center, the camera's pitch and azimuth angles are obtained. The fire center position is then calculated using the geometric relationship between the camera and the fire center.

[0082] The latitude and longitude of the camera and altitude h c is the actual installation parameter, which is a known quantity. Initially, the preset altitude of the fire center P is h p .

[0083] Step 4062, calculate the longitude and latitude coordinates of the fire center according to the preset altitude of the fire center, the altitude of the camera, the longitude and latitude of the camera, the pitch angle and the azimuth angle.

[0084] Specifically, the geometric relationship between the camera and the fire center is used to calculate the longitude and latitude corresponding to the fire center, that is, the position of the preset fire center, based on the altitude of the preset fire center, the altitude of the camera, the longitude and latitude of the camera, the pitch angle and the azimuth angle.

[0085] Specifically, step 4062, as Figure 4 As shown, including:

[0086] Step S1, determining the horizontal distance between the fire center and the camera according to the preset altitude of the fire center, the altitude and pitch angle of the camera.

[0087] The pitch angle θ is the angle between the camera's optical axis and the horizontal plane, reflecting the degree to which the camera is tilted upward or downward.

[0088] The camera's altitude h can be calculated by the pitch angle θ. c The altitude of the preset fire center h p The vertical height difference between them is converted into the horizontal distance D. Specifically, the preset horizontal distance D from the fire center to the camera is:

[0089]

[0090] Step S2: Calculate the east-west displacement and north-south displacement of the fire center based on the horizontal distance and azimuth between the fire center and the camera.

[0091] Azimuth It is the angle between the projection of the camera's optical axis on the horizontal plane and the north direction, reflecting the camera's horizontal orientation.

[0092] By azimuth The displacements Δx and Δy of the fire center in the east-west and north-south directions can be determined:

[0093]

[0094] Step S3: Convert the east-west displacement and the north-south displacement of the fire center into the latitude and longitude increments of the fire center relative to the camera.

[0095] Considering the curvature of the earth, the east-west displacement Δx and north-south displacement Δy of the preset fire situation are converted into longitude and latitude increments Δλ and

[0096]

[0097] Where R is the radius of the Earth, is the latitude of the camera. The length of the longitude changes with the latitude, so it needs to be multiplied by Corrected east-west displacement.

[0098] Step S4, obtaining the longitude and latitude coordinates of the fire center according to the longitude and latitude of the camera and the longitude and latitude increments.

[0099] Add the latitude and longitude increments to the camera's latitude and longitude to obtain the latitude and longitude of the preset fire center.

[0100] λ t =Δλc +Δλ

[0101]

[0102] In this embodiment, the position of the fire center is calculated by trigonometric functions using the geometric relationship between the camera and the fire center.

[0103] Step 4063, determine the actual altitude corresponding to the longitude and latitude coordinates.

[0104] Using the Geographic Information System (GIS), query the actual altitude ht_real corresponding to the latitude and longitude coordinates calculated based on the preset altitude of the fire center.

[0105] Step 4064: determine whether the absolute value of the error between the actual altitude and the preset altitude is greater than a threshold. If so, execute step 4065 to update the actual altitude to the preset altitude of the fire center, that is, to set the preset altitude h of the fire center to p Update to the actual altitude ht_real. After step 4065, return to execute step 4062 and subsequent steps until the absolute value of the error is less than the threshold.

[0106] The error calculation formula is:

[0107] e=ht_real-h p

[0108] If the absolute value of the error is less than the threshold, step 4066 is executed to obtain the location of the fire according to the latitude and longitude coordinates of the fire center.

[0109] The iteration stops when the absolute value of the error is less than the threshold. The longitude and latitude corresponding to the fire center at this time are the actual location of the fire center.

[0110] By combining geometric triangulation and iterative optimization, this method can gradually approximate and accurately determine the longitude and latitude of the fire center, and has high practicality and accuracy.

[0111] In one embodiment, the fire monitoring method further includes: calling a fire dispatch server to determine a rescue center according to the location of the fire; generating a rescue path from the rescue center to the location of the fire, and sending the rescue path to a rescue terminal.

[0112] The fire dispatch server's primary function is to dispatch fire rescue centers within its area based on specific principles when a fire occurs, arranging for one or more centers to respond. The server determines the fire rescue center based on various dispatching principles, including proximity and hierarchical dispatch.

[0113] The rescue terminal is a communication-capable terminal device, typically a handheld terminal used by managers or staff at a fire rescue center. The computer calls the fire dispatch server, identifies the rescue center based on the fire location, generates a rescue route from the rescue center to the fire location, and sends the route to the corresponding rescue terminal. This allows fire rescue center staff to receive dispatch instructions and the route simultaneously via the rescue terminal, enabling rapid response and precise dispatch from the fire rescue center, improving rescue efficiency.

[0114] In one embodiment, Figure 1 In the system architecture shown, at least one of real-time images, fire description information, and rescue strategies is sent to the rescue terminal 50 via the fire dispatch server 40 .

[0115] exist Figure 2 In the system architecture shown, at least one of real-time images, fire description information, and rescue strategies is sent to the rescue terminal 50 via the fire monitoring server 70 .

[0116] Rescue personnel can thus gain a timely understanding of the real-time situation at the fire scene and the rescue strategy through real-time images, fire descriptions, and rescue strategies. For example, by receiving real-time images and fire descriptions, they can obtain key information such as the fire scene layout, fire spread, and smoke concentration in advance. This understanding of the on-site situation, combined with the rescue strategy, provides a richer and more comprehensive basis for rescue decision-making. This improves rescue efficiency, enhances decision-making accuracy, and ensures rescue safety.

[0117] During the firefighting and rescue process, the computer device uses a large visual model to analyze real-time images. This real-time image can be captured by a camera monitoring the scene or by a rescue terminal. If the large visual model detects the presence of fireworks or fire in the real-time image, it generates a fire description for the real-time image using the large visual model. The large language model is then used to generate a rescue strategy based on the fire description and transmit the strategy to the rescue terminal until the large visual model detects the absence of fireworks or fire in the real-time image. In other words, after detecting a fire, the computer device continuously updates the rescue strategy based on the real-time images of the fire scene. This allows the rescue strategy to be dynamically adjusted based on the real-time and ever-changing conditions of the fire scene, ensuring that the rescue strategy is appropriate for the current situation and improving rescue efficiency and effectiveness. For example, if a fire is detected spreading rapidly in a certain area, additional firefighting forces can be dispatched to that area in a timely manner to reduce the possibility of the fire spreading. Alternatively, as the fire grows, the computer device can provide timely guidance for the fire brigade to more quickly deploy personnel, vehicles, and equipment.

[0118] During the firefighting and rescue process, the computer device can not only use the large visual model to analyze real-time images of the fire scene in real time, generate fire description information, and formulate rescue strategies accordingly, but also further enhance interaction with rescue personnel. Specifically, the computer device can receive real-time questions submitted by the rescue terminal during the rescue process and use the large language model for intelligent analysis and processing, generating detailed and accurate responses. This interactive mechanism enables rescue personnel to obtain critical information and professional guidance in a timely manner when faced with complex and changing fire scenes, allowing them to make more reasonable decisions and further improve rescue efficiency and safety.

[0119] Another aspect of the present application further provides a computer device, which can be Figure 1 The master device shown, or Figure 2 The fire monitoring server shown in FIG. The computer device includes a processor and a memory connected to the processor. The memory stores a computer program executable by the processor. When the computer program is executed by the processor, the various processes of the fire monitoring method embodiment described above are implemented, and the same technical effects are achieved. To avoid repetition, these are not described here.

[0120] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0121] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), including a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0122] The above are only specific embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A fire monitoring method, characterized in that: The method comprises: Obtain real-time images of the monitoring scene; calling the visual big model to analyze the real-time image, and generating fire description information of the real-time image through the visual big model when the visual big model detects that there is a fire target in the real-time image; A large language model is called to generate a rescue strategy based on the fire description information.

2. The method according to claim 1, characterized in that The method further comprises: Determine the location of the fire based on the camera's installation location information and the relative relationship between the camera and the center of the fire; The calling of the large language model and generating a rescue strategy based on the fire description information includes: inputting a strategy generation prompt word, the fire location and the fire description information into the large language model, the strategy generation prompt word instructing the large language model to generate a rescue strategy based on the fire location and the fire description information.

3. The method according to claim 2, characterized in that The rescue strategy includes at least one of a fire extinguishing strategy, a personnel evacuation strategy, a rescue resource deployment strategy, and a secondary disaster prevention strategy.

4. The method according to claim 3, characterized in that The fire-fighting strategy includes the type of fire-fighting equipment, the number of people dispatched, and fire-fighting tactics; the personnel evacuation strategy includes the evacuation route and evacuation priority; the rescue resource allocation strategy includes the fire-fighting resource allocation strategy and the auxiliary rescue resource allocation strategy; the secondary disaster prevention strategy includes the secondary disaster risks that may be caused, and secondary disaster protection measures.

5. The method according to claim 1 or 2, characterized in that The calling of the visual big model to analyze the real-time image, and generating fire description information of the real-time image by the visual big model when the visual big model detects that there is a fire target in the real-time image, includes: Inputting a fire identification prompt word and the real-time image into the visual big model, wherein the fire identification prompt word instructs the visual big model to detect a fire target in the real-time image; When it is determined that the fireworks target exists in the real-time image, a fire description prompt word and the real-time image are input into the visual big model, and the fire description prompt word instructs the visual big model to generate fire description information of the real-time image.

6. The method according to claim 2, characterized in that The camera installation location includes: the camera's latitude, longitude, and altitude; and determining the fire location based on the camera installation location information and the relative relationship between the camera and the fire center point includes: Controlling the camera to adjust and aim at the center of the fire, and obtaining the pitch angle and azimuth angle of the camera when it is aimed at the center of the fire, as well as the latitude, longitude and altitude of the camera; Calculating the longitude and latitude coordinates of the fire center according to the preset altitude of the fire center, the altitude of the camera, the longitude and latitude of the camera, the pitch angle and the azimuth angle; Determining the actual altitude corresponding to the latitude and longitude coordinates; When the absolute value of the error between the actual altitude and the preset altitude is greater than a threshold, the actual altitude is updated to the preset altitude of the fire center, and the process returns to the step of calculating the longitude and latitude corresponding to the fire center based on the preset altitude of the fire center, the altitude of the camera, the longitude and latitude of the camera, the pitch angle, and the azimuth, until the absolute value of the error is less than a threshold, and the location of the fire is obtained based on the longitude and latitude coordinates of the fire center.

7. The method according to claim 6, characterized in that The method of calculating the longitude and latitude coordinates of the fire center according to the preset altitude of the fire center, the altitude of the camera, the longitude and latitude of the camera, the pitch angle and the azimuth angle includes: Determining the horizontal distance between the fire center and the camera according to the preset altitude of the fire center, the altitude of the camera, and the pitch angle; Calculating the east-west displacement and north-south displacement of the fire center based on the horizontal distance and azimuth between the fire center and the camera; Converting the east-west displacement and the north-south displacement of the fire center into longitude and latitude increments of the fire center relative to the camera; The longitude and latitude coordinates of the fire center are obtained according to the longitude and latitude of the camera and the longitude and latitude increments.

8. The method according to claim 2, characterized in that The method further comprises: Calling a fire dispatch server to determine a rescue center based on the location of the fire; A rescue path from the rescue center to the fire location is generated, and the rescue path is sent to a rescue terminal corresponding to the rescue center.

9. The method according to claim 8, characterized in that The method further comprises: The real-time image, the fire description information and at least one of the rescue strategy are sent to the rescue terminal, or the real-time image, the fire description information and at least one of the rescue strategy are sent to the rescue terminal through the fire dispatch server.

10. A computer device comprising a processor and a memory connected to the processor, characterized in that: The memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, the steps of the fire monitoring method according to any one of claims 1 to 9 are implemented.

11. A fire monitoring system, characterized in that: include: A fire dispatch server, a rescue terminal, a camera set up at the monitoring site and a computer device as described in claim 10, wherein the camera is communicatively connected to the computer device, and the computer device is communicatively connected to the fire dispatch server; the rescue terminal is communicatively connected to the fire dispatch server.

12. A fire monitoring system, characterized in that: include: A fire dispatch server, a rescue terminal, a computer device as described in claim 10, and a camera and a main control terminal set up at the monitoring site, the camera is communicatively connected to the main control terminal, the main control terminal at each monitoring site is communicatively connected to the computer device, and the computer device is communicatively connected to the fire dispatch server; the rescue terminal is communicatively connected to the computer device or to the fire dispatch server.

Images