Fire alarm method and system
By using deep learning fire detection models on edge devices to detect high-risk areas of fire, the problem of fire false alarm in the prior art is solved, and higher fire alarm accuracy is achieved.
Patent Information
- Application Number
- CN202411809802.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-04-25
AI Technical Summary
There is a problem of fire false alarms in the prior art, which leads to low accuracy of fire alarms.
The fire alarm method based on edge devices is adopted. By acquiring the video data collected by the monitoring device, decoding the video, the image data is input into the deep learning-based fire detection model, the fire detection results are output, and the alarm command is sent to the alarm device based on the detection results.
It improves the accuracy of fire detection results, reduces the situation of fire false alarms, and thus improves the accuracy of fire alarms.
Smart Images
Figure CN119888951A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of fire detection, and in particular to a fire alarm method and system. Background Art
[0002] Fire is one of the most common types of emergency disasters. It is characterized by suddenness and destructiveness. It is one of the hazards to public safety. Therefore, once a fire occurs, the consequences will be disastrous.
[0003] Traditional fire alarms are usually implemented using a threshold-based central alarm method, specifically: the monitoring system collects data from the monitored object in real time and compares it with the preset threshold; when the collected data exceeds or falls below the threshold, the monitoring system triggers the alarm mechanism to achieve the purpose of fire alarm.
[0004] However, if the threshold is not set accurately, there may be a problem of false fire alarms when using the above method. Summary of the invention
[0005] The present application provides a fire alarm method and system to solve the problem of false fire alarms in the prior art, thereby effectively improving the accuracy of fire alarms.
[0006] The present application provides a fire alarm method, which is applied to an edge device, and the method includes: Obtain video data from surveillance equipment in high-risk areas for fire; Inputting the image data obtained after decoding the video data into the fire detection model, and outputting the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is trained based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels; An alarm instruction is sent to an alarm device based on the fire detection result, and the alarm instruction is used for the alarm device to perform a corresponding alarm operation.
[0007] According to a fire alarm method provided by the present application, sending an alarm instruction to an alarm device based on the fire detection result includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction; The alarm instruction is sent to the alarm device.
[0008] According to a fire alarm method provided by the present application, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, determining fire parameter information, the parameter information including flame height, smoke density, temperature and spread speed; An alarm level corresponding to the fire is determined based on the parameter information of the fire, and the alarm instruction is generated based on the alarm level.
[0009] According to a fire alarm method provided by the present application, the method further includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, superimposing a bounding box of a target location where the fire has occurred on the image data to obtain superimposed target image data; The target image data is sent to a cloud server, and the cloud server is used to display the target image data on a user interface.
[0010] According to a fire alarm method provided by the present application, the method further includes: Receive the fire detection model sent by the cloud server; wherein the fire detection model is obtained by the cloud server based on the following training method: Inputting image data samples obtained by decoding the video data samples of each of the plurality of high-risk area samples into an initial fire detection model to obtain fire detection result samples corresponding to each of the image data samples; Based on the fire detection result samples and fire detection labels corresponding to each of the image data samples, the model parameters of the initial fire detection model are updated to obtain the fire detection model.
[0011] The present application also provides a fire alarm system, comprising: an edge device, a monitoring device and an alarm device connected to a network of the edge device; Wherein, the monitoring device is used to collect video data of the high-risk fire area and send the video data to the edge device; The edge device is used to input the image data obtained after decoding the video data into the fire detection model, and output the fire detection result corresponding to the high-risk fire area through the fire detection result; and send an alarm instruction to the alarm device based on the fire detection result; wherein the fire detection model is trained based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels; The alarm device is used to perform a corresponding alarm operation based on the alarm instruction.
[0012] According to a fire alarm system provided by the present application, sending an alarm instruction to an alarm device based on the fire detection result includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction; The alarm instruction is sent to the alarm device.
[0013] According to a fire alarm system provided by the present application, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, determining fire parameter information, the parameter information including flame height, smoke density, temperature and spread speed; An alarm level corresponding to the fire is determined based on the parameter information of the fire, and the alarm instruction is generated based on the alarm level.
[0014] According to a fire alarm system provided by the present application, the system further includes a cloud server; The edge device is further configured to, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, superimpose a boundary box of a target location where the fire has occurred on the image data to obtain superimposed target image data; and send the target image data to the cloud server; The cloud server is used to display the target image data on a user interface.
[0015] According to a fire alarm system provided by the present application, the edge device is further used to receive the fire detection model sent by the cloud server; wherein the fire detection model is obtained by the cloud server based on the following training method: Inputting image data samples obtained by decoding the video data of the frequency data samples of the plurality of high-risk area samples into the initial fire detection model to obtain fire detection result samples corresponding to the video data samples; Based on the fire detection result samples and fire detection labels corresponding to each of the image data samples, the model parameters of the initial fire detection model are updated to obtain the fire detection model.
[0016] The present application also provides an edge device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the fire alarm methods described above when executing the computer program.
[0017] The present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the fire alarm method as described above is implemented.
[0018] The present application also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the fire alarm method as described above is implemented.
[0019] The fire alarm method and system provided by the present application can first obtain the video data of the high-risk fire area collected by the monitoring equipment when making a fire alarm; input the image data obtained after decoding the video data into the fire detection model, and output the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection label training; then send an alarm instruction to the alarm device based on the fire detection result, and the alarm instruction is used for the alarm device to perform the corresponding alarm operation. In this way, the fire detection of the high-risk fire area is combined with the fire detection model based on deep learning, and the fire detection result corresponding to the high-risk fire area is output, which can effectively improve the accuracy of the fire detection result, so that when the alarm instruction is subsequently sent to the alarm device based on the fire detection result, the problem of false fire alarm in the prior art can be solved, thereby effectively improving the accuracy of the fire alarm. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0021] Figure 1 A flow chart of a fire alarm method provided in an embodiment of the present application.
[0022] Figure 2 A flowchart of a method for sending an alarm instruction to an alarm device based on a fire detection result, wherein the alarm instruction is used by the alarm device to perform a corresponding alarm operation, is provided in an embodiment of the present application.
[0023] Figure 3 A schematic diagram of a fire alarm method provided in an embodiment of the present application.
[0024] Figure 4 A flow chart of a fire alarm system provided in an embodiment of the present application.
[0025] Figure 5 A schematic diagram of the physical structure of an edge device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0026] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0027] In the embodiments of the present application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of the present application, the character " / " generally indicates that the associated objects before and after are in an "or" relationship.
[0028] The technical solution provided by the embodiment of the present application can be applied to fire detection scenarios. Traditional fire alarms are usually implemented using a central alarm method based on a threshold value, but false fire alarms may occur due to factors such as environmental noise.
[0029] In order to solve the problem of false fire alarms in the prior art and thereby improve the accuracy of fire alarms, an embodiment of the present application provides a fire alarm method that can be applied to a fire alarm system. The fire alarm system may include an edge device, a monitoring device and an alarm device connected to a network of the edge device, and of course, may also include a cloud server, etc., which may be specifically configured according to actual needs.
[0030] Among them, the monitoring equipment is used to collect video data of high-risk fire areas and send the video data to the edge device. The edge device is used to input the image data obtained after decoding the video data into the fire detection model, and output the fire detection results corresponding to the high-risk fire areas through the fire detection results; and send an alarm instruction to the alarm device based on the fire detection results. The alarm device is used to perform corresponding alarm operations based on the alarm instruction. In this way, combining the fire detection model based on deep learning to perform fire detection on high-risk fire areas and output the fire detection results corresponding to the high-risk fire areas can effectively improve the accuracy of the fire detection results, so that when the alarm instruction is subsequently sent to the alarm device based on the fire detection results, the accuracy of the fire alarm can be effectively improved.
[0031] It can be understood that the execution subject of the fire alarm method provided in the embodiment of the present application can be the above-mentioned edge device, or it can be a fire alarm device arranged in the edge device, and the fire alarm device can be implemented by software, hardware or a combination of both.
[0032] The fire alarm method provided by the present application will be described in detail below through the following specific embodiments. It is understandable that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.
[0033] Figure 1 A flow chart of a fire alarm method provided in an embodiment of the present application is applied to an edge device, which can be deployed in a high-risk fire area, and a fire detection model for fire detection is deployed in the edge device. For example, see Figure 1 As shown, the fire alarm method may include: S101. Obtain video data of high-risk fire areas collected by monitoring equipment.
[0034] For example, high-risk fire areas, such as industrial plants, warehouses, chemical plants, etc., are usually used to store a large number of flammable and explosive items.
[0035] For example, the monitoring device may be a gun-type camera or other types of network cameras, and may be configured according to actual needs.
[0036] When collecting video data of high-risk fire areas through control equipment, monitoring equipment can be set up in advance according to the on-site environment of the high-risk fire areas to ensure that the high-risk fire areas can be clearly photographed within the monitoring range of the monitoring equipment; and it is necessary to ensure that the communication link between the monitoring equipment and the edge device is unobstructed.
[0037] In this way, after the monitoring device collects video data of the high-risk fire area, the collected video data can be sent to the edge device, so that the edge device can perform fire detection on the high-risk fire area based on the video data, that is, execute the following S102.
[0038] S102. Input the image data obtained after decoding the video data into the fire detection model, and output the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels for training.
[0039] For example, when decoding the video data, the development toolkit of the monitoring device can decode and callback the YUV format image data of the monitoring device, and then use the FFMPEG tool to decode the video data and convert it into BGR (Blue, Green, Red) format image data.
[0040] For example, the fire detection model may be a YOLOv7 model based on deep learning, or other models based on deep learning, such as a convolutional neural network (CNN) model based on deep learning, etc., and may be specifically set according to actual needs.
[0041] Generally, after obtaining the image data obtained after decoding the video data, if the image quality of the image data is good, the image data can be directly used as the input of the fire detection model to output the fire detection results corresponding to the high-risk fire areas through the fire detection model; on the contrary, if the image quality of the image data is poor, the image data can be preprocessed first, such as denoising, image enhancement, geometric correction, atmospheric correction, etc., and the preprocessed image data can be used as the input of the fire detection model to output the fire detection results corresponding to the high-risk fire areas through the fire detection model.
[0042] For example, in the embodiment of the present application, the fire detection result corresponding to the high-risk fire area outputted by the fire detection result may include the occurrence of a fire and the confidence level of the occurrence of a fire, or the fire detection result may include the non-occurrence of a fire and the confidence level of the non-occurrence of a fire. It is worth noting that when an open flame and / or smoke appears in a high-risk fire area, it can be understood that a fire has occurred in the high-risk fire area.
[0043] S103. Send an alarm instruction to the alarm device based on the fire detection result, and the alarm instruction is used by the alarm device to perform a corresponding alarm operation.
[0044] For example, the alarm device may be an audible and visual alarm, and the corresponding alarm operation may be a loud alarm and flashing lights, so that the personnel near the high-risk fire area can be reminded to evacuate quickly through the alarm operation, thereby effectively improving the safety of the personnel. In addition, text messages can be sent to the relevant persons in charge through the Internet to ensure that they are informed of the fire situation as soon as possible, effectively improving the fire emergency response capability, thereby ensuring the safety of personnel and property.
[0045] It can be seen that in the embodiment of the present application, when a fire alarm is performed, the video data of the high-risk fire area collected by the monitoring device can be obtained first; the image data obtained after the video data is decoded is input into the fire detection model, and the fire detection result corresponding to the high-risk fire area is output through the fire detection result; wherein, the fire detection model is based on the image data samples obtained after the video data samples of each of the multiple high-risk area samples are decoded, and the corresponding fire detection labels are trained; then, based on the fire detection result, an alarm instruction is sent to the alarm device, and the alarm instruction is used for the alarm device to perform the corresponding alarm operation. In this way, the fire detection of the high-risk fire area is combined with the fire detection model based on deep learning, and the fire detection result corresponding to the high-risk fire area is output, which can effectively improve the accuracy of the fire detection result, so that when the alarm instruction is subsequently sent to the alarm device based on the fire detection result, the problem of false fire alarm in the prior art can be solved, thereby effectively improving the accuracy of the fire alarm.
[0046] Based on the above Figure 1 In the embodiment shown, in the above 103, an alarm instruction is sent to the alarm device based on the fire detection result. When the alarm instruction is used for the alarm device to perform the corresponding alarm operation, if the fire detection result indicates that no fire has occurred in the high-risk fire area, and the confidence that no fire has occurred is greater than a preset confidence threshold, it means that the high-risk fire area is currently in a safe state, and there is no need to send an alarm instruction to the alarm device; on the contrary, if the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence that a fire has occurred is greater than a preset confidence threshold, it means that the high-risk fire area is currently in a dangerous state, and an alarm instruction is sent to the alarm device. For details, please refer to the following Figure 2 The illustrated embodiments are described in detail.
[0047] Figure 2 A flowchart of a method for sending an alarm instruction to an alarm device based on a fire detection result provided in an embodiment of the present application, wherein the alarm instruction is used by the alarm device to perform a corresponding alarm operation, the method may include: S201. When a fire detection result indicates that a fire has occurred in a high-risk area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, an alarm instruction is generated.
[0048] The value of the preset confidence threshold can be set according to actual needs. Here, the embodiment of the present application does not further limit the specific value of the preset confidence threshold.
[0049] For example, in an embodiment of the present application, when the fire detection results indicate that a fire has occurred in a high-risk area for fire, and the confidence level of the fire occurrence is greater than a preset confidence threshold, when an alarm instruction is generated, parameter information of the fire can be determined, and the parameter information includes flame height, smoke concentration, temperature and spread speed; and the alarm level corresponding to the fire is determined based on the parameter information of the fire, and an alarm instruction is generated based on the alarm level. In this way, by determining the alarm level corresponding to the fire based on the parameter information of the fire, alarm instructions of different levels can be generated in a targeted manner, which can provide an accurate basis for subsequent fire emergency response.
[0050] For example, the alarm levels corresponding to fire can be divided into level 1 alarm level, level 2 alarm level and level 3 alarm level. The lower the alarm level, the more serious the corresponding fire situation is, and the more urgent the corresponding generated alarm instruction is.
[0051] For example, when determining the alarm level corresponding to a fire based on the parameter information of the fire, the weights of the flame height, smoke concentration, temperature and spread speed can be determined respectively, and based on the weights of the flame height, smoke concentration, temperature and spread speed, the flame height, smoke concentration, temperature and spread speed are weighted to obtain a target alarm value; and the target alarm value is compared with the alarm thresholds corresponding to different alarm levels. Assuming that the alarm thresholds a and b are preset, if the target alarm value is greater than or equal to the alarm threshold a, it can be determined that the alarm level corresponding to the fire is the first alarm level; if the target alarm value is greater than or equal to the alarm threshold b, and less than the alarm threshold a, it can be determined that the alarm level corresponding to the fire is the second alarm level; if the target alarm value is less than the alarm threshold b, it can be determined that the alarm level corresponding to the fire is the third alarm level.
[0052] For example, the alarm instructions corresponding to the first-level alarm level may include: immediately starting the emergency evacuation procedure and notifying all personnel to evacuate to a safe area quickly; at the same time, starting the fire-fighting system, including automatic sprinkler fire-extinguishing system, gas fire-extinguishing system, etc., to carry out initial fire-fighting.
[0053] The alarm instructions corresponding to the second-level alarm level may include: immediately organizing personnel to carry out initial fire extinguishing and starting some fire-fighting systems; at the same time, notifying relevant personnel to prepare for evacuation and be ready to evacuate at any time.
[0054] The alarm instructions corresponding to the third level alarm level may include: strengthening fire monitoring and increasing vigilance; at the same time, notifying relevant personnel to investigate and deal with fire hazards.
[0055] After the alarm instruction is generated, the alarm instruction can be sent to the alarm device, that is, the following S202 is executed.
[0056] S202: Send an alarm instruction to an alarm device.
[0057] Among them, the alarm instruction is used for the alarm device to perform corresponding alarm operations.
[0058] Combined with the description in S201 above, for the alarm instruction corresponding to the first alarm level, the corresponding alarm operation may include: sending out an emergency alarm sound (similar to the sound of a 119 fire truck) through the alarm device, and lighting up the red fire alarm indicator light. In addition, the time, location, and on-site pictures of the alarm can be recorded, and the alarm instruction can be sent to relevant personnel through various methods such as the broadcasting system, mobile phone text messages, and emails.
[0059] For the alarm instructions corresponding to the second-level alarm level, the corresponding alarm operations may include: issuing an important alarm sound (different from the emergency alarm sound, but also attracting attention) through the alarm device, lighting up the red fire alarm indicator light (or combining with other color indicators for distinction). In addition, the time, location, and on-site pictures of the alarm can be recorded, and alarm instructions can be issued to relevant personnel through the broadcasting system, internal communication system, etc.
[0060] For the alarm instructions corresponding to the third level of alarm, the corresponding alarm operations may include: issuing a general alarm sound (relatively soft, but enough to attract attention) through the alarm device, lighting up the corresponding indicator light (such as a yellow warning light). In addition, the time, location, and on-site pictures of the alarm can be recorded, and the alarm instructions can be issued to relevant personnel through the internal communication system, email, etc.
[0061] Combined with the above description, when the fire detection results indicate that a fire has occurred in a high-risk area, and the confidence level of the fire is greater than a preset confidence threshold, an alarm instruction is generated; and an alarm instruction is sent to the alarm device, and the alarm instruction is used by the alarm device to perform the corresponding alarm operation. In this way, combined with the fire detection model based on deep learning, after detecting a fire in a high-risk area, an alarm instruction can be sent to the alarm device in a timely manner, which can not only solve the problem of false fire alarms in the prior art and effectively improve the accuracy of fire alarms; but also through the alarm instruction, the relevant personnel can be prompted to perform the corresponding fire emergency response, thereby effectively improving the fire emergency response capability and ensuring the safety of personnel and property.
[0062] For example, in an embodiment of the present application, the image data obtained after decoding the collected video data of the high-risk area and the corresponding fire detection results can also be uploaded to the cloud server so that the cloud server can train and optimize the fire detection model to improve the accuracy of the fire detection model detection, thereby providing strong protection for fire safety.
[0063] Based on any of the above embodiments, for example, when the fire detection results indicate that a fire has occurred in a high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, the bounding box of the target location where the fire has occurred can be superimposed on the image data to obtain superimposed target image data; and the target image data can be sent to a cloud server, which is used to display the target image data on a user interface. By displaying the target image data on the user interface, relevant personnel can intuitively understand the situation in the high-risk fire area based on the displayed target image data, thereby providing a basis for the relevant personnel to subsequently perform corresponding fire emergency responses.
[0064] For example, the cloud server may be a graphics processing unit server (Graphics Processing Unit Server, GPU).
[0065] Based on any of the above embodiments, considering that the computing power of the edge device is usually limited, for example, in the embodiment of the present application, the edge device can receive a fire detection model sent by the cloud server; wherein the fire detection model is obtained by the cloud server based on the following training method: The image data samples obtained after video data decoding of each of the multiple high-risk area samples are input into the initial fire detection model to obtain the fire detection result samples corresponding to each image data sample; and based on the fire detection result samples and fire detection labels corresponding to each image data sample, the model parameters of the initial fire detection model are updated to obtain the fire detection model.
[0066] Generally, when the initial fire detection model is trained based on the image data samples corresponding to the video data samples of multiple high-risk area samples, if the image quality of the image data samples is good, the initial fire detection model can be directly trained based on the image data samples corresponding to the video data samples of multiple high-risk area samples; on the contrary, if the image quality of the image data samples is poor, the image data samples can be preprocessed first, such as denoising, image enhancement, geometric correction, atmospheric correction, etc., and the initial fire detection model can be trained based on the preprocessed image data samples.
[0067] For example, when the model parameters of the initial fire detection model are updated based on the fire detection result samples and fire detection labels corresponding to each image data sample, for each image data sample, a loss function corresponding to the image data sample can be constructed based on the fire detection result samples and fire detection labels corresponding to the image data sample; and the loss function corresponding to each image data sample is weighted to obtain a target loss function, so as to iteratively update the model parameters of the initial fire detection model based on the target loss function to obtain a fire detection model.
[0068] It is understandable that the process of the above-mentioned fire detection model can be carried out on the cloud server side. For the cloud server, the environment can be prepared, necessary dependencies (such as Python, PyTorch, etc.) can be installed, and a suitable initial fire detection model can be selected as the basis of the fire detection model. For example, in the embodiment of the present application, the initial fire detection model can select the YOLOv7 model based on deep learning, and based on the fire detection result samples and fire detection labels corresponding to each image data sample, the model parameters of the initial fire detection model are updated, including updating the weights and biases of the initial fire detection model to train the final fire detection model. In addition, the initial fire detection model can also be continuously iterated in combination with the image data samples corresponding to the video data samples of each new high-risk area sample to improve the accuracy of the fire detection model detection, thereby providing a strong guarantee for fire safety.
[0069] To facilitate understanding of the fire alarm method provided in the embodiments of the present application, for example, see Figure 3 As shown, Figure 3 A framework diagram of a fire alarm method provided in an embodiment of the present application, wherein the cloud server can pre-select a suitable initial fire detection model as the basis of the fire detection model, and based on the fire detection result samples and fire detection labels corresponding to each image data sample, update the model parameters of the initial fire detection model, including updating the weights and biases of the initial fire detection model, so as to train the final fire detection model, and send the trained fire detection model to the edge device.
[0070] After collecting video data from high-risk fire areas, the monitoring equipment sends the collected video data to the edge device; the edge device decodes the video data to obtain corresponding image data, and inputs the image data into the fire detection model to determine whether a fire is detected, and if the confidence level of the fire is greater than a preset confidence threshold, an alarm instruction is generated; and an alarm instruction is sent to the alarm device. In this way, fire detection is performed on high-risk fire areas in combination with a fire detection model based on deep learning, and the fire detection results corresponding to the high-risk fire areas are output, which can effectively improve the accuracy of the fire detection results, so that when an alarm instruction is subsequently sent to the alarm device based on the fire detection results, the problem of false fire alarms in the prior art can be solved, thereby effectively improving the accuracy of fire alarms.
[0071] The fire alarm system provided by the present application is described below. The fire alarm system described below and the fire alarm method described above can be referenced to each other.
[0072] Figure 4 A flow chart of a fire alarm system provided in an embodiment of the present application is provided. For example, see Figure 4 As shown, the fire alarm system 40 may include: an edge device 401 , a monitoring device 402 and an alarm device 403 connected to the edge device 401 through a network.
[0073] The monitoring device 402 is used to collect video data of the high-risk fire area and send the video data to the edge device 401; The edge device 401 is used to input the image data obtained after decoding the video data into the fire detection model, and output the fire detection result corresponding to the high-risk fire area through the fire detection result; and send an alarm instruction to the alarm device 403 based on the fire detection result; wherein the fire detection model is based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels. The alarm device 403 is used to perform a corresponding alarm operation based on the alarm instruction.
[0074] For example, in the embodiment of the present application, sending an alarm instruction to the alarm device 403 based on the fire detection result includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction; The alarm instruction is sent to the alarm device 403.
[0075] For example, in an embodiment of the present application, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, determining fire parameter information, the parameter information including flame height, smoke density, temperature and spread speed; An alarm level corresponding to the fire is determined based on the parameter information of the fire, and the alarm instruction is generated based on the alarm level.
[0076] For example, in the embodiment of the present application, it is characterized in that the system 40 further includes a cloud server 404; The edge device 401 is further configured to, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, superimpose a boundary box of the target location where the fire has occurred on the image data to obtain superimposed target image data; and send the target image data to the cloud server 404; The cloud server 404 is used to display the target image data on a user interface.
[0077] For example, in the embodiment of the present application, the edge device 401 is further used to receive the fire detection model sent by the cloud server 404; wherein the fire detection model is obtained by the cloud server 404 based on the following training method: Inputting image data samples obtained by decoding the video data of the frequency data samples of the plurality of high-risk area samples into the initial fire detection model to obtain fire detection result samples corresponding to the video data samples; Based on the fire detection result samples and fire detection labels corresponding to each of the image data samples, the model parameters of the initial fire detection model are updated to obtain the fire detection model.
[0078] The fire alarm system 40 provided in the embodiment of the present application can execute the technical solution of the fire alarm method in any of the above-mentioned embodiments. Its implementation principle and beneficial effects are similar to those of the fire alarm method. Please refer to the implementation principle and beneficial effects of the fire alarm method, and no further details will be given here.
[0079] Figure 5 A schematic diagram of the physical structure of an edge device provided in an embodiment of the present application, such as Figure 5As shown, the edge device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the fire alarm method, which includes: obtaining video data of a high-risk fire area collected by a monitoring device; inputting image data obtained after video data decoding of the video data into a fire detection model, and outputting the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is obtained by training the image data samples obtained after video data decoding of the video data samples of each of the multiple high-risk area samples and the corresponding fire detection labels; sending an alarm instruction to the alarm device based on the fire detection result, and the alarm instruction is used for the alarm device to perform a corresponding alarm operation.
[0080] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0081] On the other hand, the present application also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the fire alarm method provided by the above-mentioned methods, and the method includes: obtaining video data of a high-risk fire area collected by a monitoring device; inputting image data obtained after decoding the video data into a fire detection model, and outputting the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is based on image data samples obtained after decoding the video data samples of each of multiple high-risk area samples, and corresponding fire detection labels are trained; based on the fire detection result, an alarm instruction is sent to the alarm device, and the alarm instruction is used for the alarm device to perform a corresponding alarm operation.
[0082] On the other hand, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the fire alarm method provided by the above-mentioned methods, the method comprising: obtaining video data of a high-risk fire area collected by a monitoring device; inputting image data obtained after decoding the video data into a fire detection model, and outputting a fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is trained based on image data samples obtained after decoding the video data samples of each of a plurality of high-risk area samples, and corresponding fire detection labels; and sending an alarm instruction to an alarm device based on the fire detection result, and the alarm instruction is used for the alarm device to perform a corresponding alarm operation.
[0083] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.
[0084] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. A fire alarm method, characterized in that: Applied to an edge device, the method comprises: Obtain video data from surveillance equipment in high-risk areas for fire; Inputting the image data obtained after decoding the video data into the fire detection model, and outputting the fire detection result corresponding to the high-risk fire area through the fire detection result; wherein the fire detection model is trained based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels; An alarm instruction is sent to an alarm device based on the fire detection result, and the alarm instruction is used for the alarm device to perform a corresponding alarm operation.
2. The fire alarm method according to claim 1, characterized in that: The sending an alarm instruction to an alarm device based on the fire detection result comprises: When the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction; The alarm instruction is sent to the alarm device.
3. The fire alarm method according to claim 2, characterized in that: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, determining fire parameter information, the parameter information including flame height, smoke density, temperature and spread speed; An alarm level corresponding to the fire is determined based on the parameter information of the fire, and the alarm instruction is generated based on the alarm level.
4. The fire alarm method according to claim 1, characterized in that: The method further comprises: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, superimposing a bounding box of a target location where the fire has occurred on the image data to obtain superimposed target image data; The target image data is sent to a cloud server, and the cloud server is used to display the target image data on a user interface.
5. The fire alarm method according to any one of claims 1 to 4, characterized in that: The method further comprises: Receive the fire detection model sent by the cloud server; wherein the fire detection model is obtained by the cloud server based on the following training method: Inputting image data samples obtained by decoding the video data samples of each of the plurality of high-risk area samples into an initial fire detection model to obtain fire detection result samples corresponding to each of the image data samples; Based on the fire detection result samples and fire detection labels corresponding to each of the image data samples, the model parameters of the initial fire detection model are updated to obtain the fire detection model.
6. A fire alarm system, characterized in that: include: Edge devices, monitoring devices and alarm devices connected to the edge device network; Wherein, the monitoring device is used to collect video data of the high-risk fire area and send the video data to the edge device; The edge device is used to input the image data obtained after decoding the video data into the fire detection model, and output the fire detection result corresponding to the high-risk fire area through the fire detection result; and send an alarm instruction to the alarm device based on the fire detection result; wherein the fire detection model is trained based on the image data samples obtained after decoding the video data samples of each of the multiple high-risk area samples, and the corresponding fire detection labels; The alarm device is used to perform a corresponding alarm operation based on the alarm instruction.
7. The fire alarm system according to claim 6, characterized in that: The sending an alarm instruction to an alarm device based on the fire detection result comprises: When the fire detection result indicates that a fire has occurred in the high-risk fire area, and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction; The alarm instruction is sent to the alarm device.
8. The fire alarm system according to claim 7, characterized in that: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, generating the alarm instruction includes: When the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, determining fire parameter information, the parameter information including flame height, smoke density, temperature and spread speed; An alarm level corresponding to the fire is determined based on the parameter information of the fire, and the alarm instruction is generated based on the alarm level.
9. The fire alarm system according to any one of claims 6 to 8, characterized in that: The system also includes a cloud server; The edge device is further configured to, when the fire detection result indicates that a fire has occurred in the high-risk fire area and the confidence level of the fire occurrence is greater than a preset confidence threshold, superimpose a bounding box of a target location where the fire has occurred on the image data to obtain superimposed target image data; and sending the target image data to the cloud server; The cloud server is used to display the target image data on a user interface.
10. The fire alarm system according to claim 9, characterized in that: The edge device is further used to receive the fire detection model sent by the cloud server; wherein the fire detection model is obtained by the cloud server based on the following training method: Inputting image data samples obtained by decoding the video data of the frequency data samples of the plurality of high-risk area samples into the initial fire detection model to obtain fire detection result samples corresponding to the video data samples; Based on the fire detection result samples and fire detection labels corresponding to each of the image data samples, the model parameters of the initial fire detection model are updated to obtain the fire detection model.