A method and system for detecting early fire conditions

Abstract

The present invention relates to a detection method and system for early-stage fires, and belongs to the field of image processing technology. The method comprises: obtaining an image group captured by a camera of an area to be detected, and forming a target tensor; obtaining first mapping data and second mapping data; performing Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix; inputting the first mapping data and the second mapping data into a convolutional network respectively, and fusing the output result with the core tensor, the first factor matrix, and the second factor matrix to obtain a target fusion feature; and using a detection head to detect the target fusion feature to detect and locate the fire point. The present application provides an early perception detection for early-stage fire points, which can reduce misjudgment of fire conditions, and can effectively identify small fire points, making it easier for personnel to respond and handle them in a timely manner. Accurate positioning of the fire point can be achieved by adding an image detection head.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an early-stage fire detection method and system. Background Art

[0002] Traditional firefighting equipment is mostly passive and can only take effect after a fire breaks out, making it ineffective for fire prevention and control. With the continuous development of electronic information technology, some smart fire control systems with automatic control and autonomous monitoring functions have begun to emerge. Leveraging the development of modern information technology, smart fire control systems combine sensors, monitoring equipment, data processing, and communication technologies to achieve real-time monitoring, early warning, and control of fires.

[0003] Existing intelligent fire control systems mainly collect, process, and analyze data through interface services connected to firefighting equipment, network equipment, and other devices. However, research has found that although existing control systems have certain autonomous detection and processing capabilities, they are not accurate in handling fire situations, mainly manifested in the following aspects:

[0004] (1) Currently, most monitoring systems rely on smoke detection for identification and alarm. Due to the fluidity of smoke, multiple smoke alarms may be triggered, leading to multiple fire sprinklers operating to extinguish the fire. However, sometimes, the detection of smoke does not necessarily mean a fire is present, such as in water vapor, steam, or high-temperature environments. In addition, when an open flame appears in an office, but the diameter of the open flame is less than 10 cm and there is little smoke, the existing smoke monitoring system cannot detect the occurrence of the fire in advance, and thus cannot notify the fire chief in time to handle the fire.

[0005] (2) Fire monitoring through the interfaces of fire-fighting equipment, network equipment, etc. cannot quickly locate the fire point. It requires manual investigation or the spread of the fire triggering the smoke alarm to know the location. This is fatal for accidents such as fires where every second counts.

[0006] It can be seen that the current fire monitoring method has defects such as misjudgment of fire conditions, inability to detect in advance when the open flame is small, and the need for manual positioning of the fire point. Summary of the Invention

[0007] In order to solve the above-mentioned problems in the prior art, the present invention provides a method and system for detecting early fire conditions.

[0008] In the first aspect, an embodiment of the present application provides a method for detecting early fires, comprising: obtaining an image group taken by a camera of an area to be detected, and forming a target tensor; wherein the image group includes multiple frames of images, and each frame of image includes an identified image background and a detection object; obtaining first mapping data and second mapping data; wherein the first mapping data is a mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is a mapping of all detection objects in the image group in the target tensor; performing Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix; respectively inputting the first mapping data and the second mapping data into a convolutional network, and fusing the output results with the core tensor, the first factor matrix, and the second factor matrix to obtain a target fusion feature; and using a detection head to detect the target fusion feature to detect and locate the fire point.

[0009] Optionally, the first mapping data and the second mapping data are respectively input into the convolutional network, and the output result is fused with the core tensor, the first factor matrix and the second factor matrix to obtain the target fusion feature, including: inputting the first mapping data into the convolutional network to obtain the first feature; inputting the second mapping data into the convolutional network to obtain the second feature; wherein the output result includes the first feature and the second feature; fusing the second feature with the core tensor, the first factor matrix and the second factor matrix to obtain the first fusion feature; adding the first fusion feature to the first feature to obtain the target fusion feature.

[0010] Optionally, when the scales of the first fused feature and the first feature are not uniform, the method further includes: performing an edge-filling operation of zero values on the first fused feature and a feature with a smaller scale in the first feature.

[0011] Optionally, the method also includes: if a fire point is detected, determining the target fire-fighting equipment closest to the fire point based on the positioning data of the fire point, and uploading the detection data to the control terminal; wherein, the target fire-fighting equipment is the fire-fighting equipment closest to the fire point; wherein, the detection data includes the approval control instructions for the target fire-fighting equipment and the positioning data of the fire point.

[0012] Optionally, the convolutional network includes three convolutional layers.

[0013] Optionally, the method further includes: obtaining a first image of the area to be detected; inputting the first image into an improved CSPNet model to obtain a first feature map; wherein the improved CSPNet model adopts a convolution module based on regional attention; and using the detection head to detect the first feature map to detect smoke.

[0014] Optionally, the method further includes: obtaining all cameras in the coverage area; wherein each camera captures an area to be detected; constructing a fire warning map based on all cameras in the coverage area; wherein each camera is used as a node in the fire warning map, and an edge is established between two nodes connected in the area; based on the fire warning map, predicting the direction of fire spread.

[0015] Optionally, the edges in the fire warning graph are provided with weights; the weights of the edges are determined by the connectivity type between two nodes, the physical distance between the two nodes, and / or the flammability properties of the materials between the two nodes.

[0016] Optionally, predicting the direction of fire spread based on the fire warning map includes: when a target node with a fire is detected, determining the direction of fire spread based on the weights of the edges between other nodes adjacent to the target node in the fire warning map.

[0017] In the second aspect, the present application provides a detection system for early fire, including: a first acquisition module, used to acquire an image group taken by a camera of the area to be detected, and form a target tensor; wherein the image group includes multiple frames of images, and each frame of image includes an identified image background and a detection object; a second acquisition module, used to acquire first mapping data and second mapping data; wherein the first mapping data is the mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is the mapping of all detection objects in the image group in the target tensor; a first processing module, used to perform Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix and a second factor matrix; a second processing module, used to input the first mapping data and the second mapping data into a convolutional network respectively, and fuse the output results with the core tensor, the first factor matrix and the second factor matrix to obtain a target fusion feature; a detection module, used to detect the target fusion feature using a detection head to detect and locate the fire point.

[0018] The beneficial effects of the present invention include: This application provides a method for early fire detection. Considering that current fire warnings, primarily based on smoke recognition, are subject to misjudgment and inability to detect small fires in advance, a dynamic image-based recognition method is proposed. The core of this method lies in "fire point detection based on dynamic background modeling and multi-feature fusion based on Tucker decomposition." Specifically, a first time series mapping data is constructed based on the image background. Features are then extracted through Tucker decomposition. The decomposed background features are more compact, better able to distinguish between real fire points and background interference (such as reflective points and movable objects in an office). The decomposed features are then fused with the output of a convolutional network to form a target fusion feature. This enhances the saliency of fire point features, suppresses background redundant information, and improves the accuracy of fire point detection. This method can effectively identify small fire points (specifically, those smaller than 10 cm). In other words, this application provides early detection of fire points, reducing misjudgments of fires and effectively identifying small fire points, facilitating timely response and treatment. Furthermore, the image-based detection head enables accurate location of fire points. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of the steps of an early fire detection method provided by an embodiment of the present invention;

[0020] Figure 2 A flowchart of another method for detecting early-stage fires provided by an embodiment of the present invention;

[0021] Figure 3 A schematic diagram of a data processing process provided by an embodiment of the present invention;

[0022] Figure 4 A schematic diagram of an improved CSPNet model provided by an embodiment of the present invention;

[0023] Figure 5 A flowchart of another method for detecting early-stage fires provided by an embodiment of the present invention;

[0024] Figure 6 A module block diagram of an early fire detection system provided by an embodiment of the present invention;

[0025] Figure 7 This is a module block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In the following description, specific details such as specific system structures and technologies are provided for illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obstructing the description of the present application with unnecessary details.

[0027] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0028] Existing intelligent fire control systems mainly collect, process, and analyze data through interface services connected to firefighting equipment, network equipment, and other devices. However, research has found that although existing control systems have certain autonomous detection and processing capabilities, they are not accurate in handling fire situations, mainly manifested in the following aspects:

[0029] (1) Currently, most monitoring systems rely on smoke detection for identification and alarm. Due to the fluidity of smoke, multiple smoke alarms may be triggered, leading to multiple fire sprinklers operating to extinguish the fire. However, sometimes, the detection of smoke does not necessarily mean a fire is present, such as in water vapor, steam, or high-temperature environments. In addition, when an open flame appears in an office, but the diameter of the open flame is less than 10 cm and there is little smoke, the existing smoke monitoring system cannot detect the occurrence of the fire in advance, and thus cannot notify the fire chief in time to handle the fire.

[0030] (2) Fire monitoring through the interfaces of fire-fighting equipment, network equipment, etc. cannot quickly locate the fire point. It requires manual investigation or the spread of the fire triggering the smoke alarm to know the location. This is fatal for accidents such as fires where every second counts.

[0031] It can be seen that the current fire monitoring method has defects such as misjudgment of fire conditions, inability to detect in advance when the open flame is small, and the need for manual positioning of the fire point.

[0032] In view of the above problems, the present application proposes the following embodiments to solve the above technical problems.

[0033] See also Figure 1 , an embodiment of the present application provides a method for detecting early fire conditions, the method comprising: steps 101 to 105.

[0034] Step 101: Obtain a group of images captured by a camera of the area to be detected and form a target tensor.

[0035] The image group includes multiple frames of images, and each frame of image includes an identified image background and a detection object.

[0036] In the embodiment of the present application, the camera can be a home camera, a network camera, a monitor, or other devices with a camera function. The area to be detected is the scene area that the camera can capture, and the area to be detected can specifically be a company office, corridor, living room, computer room, etc.

[0037] Specifically, assuming that the image captured by the camera is , each image can be composed of two parts, namely ;in, Represents the image background, Indicates the detection object. The number of image background and detection object can be multiple.

[0038] The above-mentioned image background and detection object can be obtained through a top-down pixel aggregation algorithm, and of course can also be determined through any other recognition algorithm.

[0039] In this embodiment, the detection object may be a dynamic object in the image.

[0040] Specifically, the image group I within the unit time frame rate T can be taken ( , ,..., ). If T=30, then image group I includes 15 frames of images.

[0041] In this embodiment, the image frame I may be formed into a frame with a size of 1920×1080×3× The target tensor .

[0042] Step 102: Obtain first mapping data and second mapping data.

[0043] The first mapping data is the mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is the mapping of all detection objects in the image group in the target tensor.

[0044] Specifically, the first mapping data can be expressed as ,Right now, In the target tensor The mapping in ; The second mapping data can be expressed as ,Right now, In the target tensor The mapping in .

[0045] Step 103: Perform Tucker decomposition on the first mapped data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix.

[0046] That is, for the first mapping data Use Tucker decomposition to get the core tensor and two orthogonal first factor matrices And the second factor matrix .

[0047] The decomposition process can be specifically expressed as follows:

[0048] ;

[0049] in, and are regularization parameters, The value of is 2, represents the norm, represents the Frobenius norm.

[0050] Step 104: Input the first mapping data and the second mapping data into the convolutional network respectively, and fuse the output results with the core tensor, the first factor matrix and the second factor matrix to obtain the target fusion feature.

[0051] Then, the data is fused to form the target fusion features.

[0052] Step 105: Use a detection head to detect the target fusion features to detect and locate the fire point.

[0053] The detection head is a module in the neural network specifically responsible for predicting the target location and category. In this step, the target fusion features are input into the detection head to complete the detection and location of the fire point.

[0054] In summary, the embodiments of the present application provide a method for early fire detection. Considering that current fire warnings, primarily based on smoke recognition, are subject to misjudgment and incapable of early detection of small fires, an image-based dynamic recognition method is proposed. The core of this method lies in "fire point detection based on dynamic background modeling and multi-feature fusion based on Tucker decomposition." Specifically, a first mapping data set of time series is constructed based on the image background. Features are then extracted through Tucker decomposition. The decomposed background features are more compact, better able to distinguish between real fire points and background interference (such as reflective points and movable objects in an office). The decomposed features are then fused with the output of the convolutional network to form a target fusion feature. This enhances the saliency of fire point features, suppresses background redundant information, and improves the accuracy of fire point detection. This method can effectively identify small fire points (specifically, those smaller than 10 cm). In other words, the embodiments of the present application provide early detection of fire points, reducing misjudgments of fires and effectively identifying small fire points, facilitating timely response and treatment. Furthermore, the image-based detection head enables accurate location of fire points.

[0055] See also Figure 2 The above steps input the first mapping data and the second mapping data into the convolutional network respectively, and fuse the output results with the core tensor, the first factor matrix and the second factor matrix to obtain the target fusion feature, which can specifically include: steps 201 to 204.

[0056] Step 201: Input the first mapping data into the convolutional network to obtain the first feature.

[0057] Specifically, the first mapping data Input into the convolutional network to get the first feature .

[0058] Step 202: Input the second mapping data into the convolutional network to obtain the second feature.

[0059] Specifically, the second mapping data Input into the convolutional network to obtain the second feature .

[0060] Among them, the output results include the first feature and the second feature .

[0061] Step 203: Fuse the second feature with the core tensor, the first factor matrix, and the second factor matrix to obtain a first fused feature.

[0062] Specifically, the fusion method can refer to the following formula:

[0063] ;

[0064] in, represents the first fusion feature, represents the hyperparameter, represents the L1 norm, Represents the natural logarithm of the determinant of the matrix. The remaining parameters can be referred to the explanations in the previous formula and are not repeated here.

[0065] It should be noted that, in the embodiment of the present application, the information content of the determinant logarithm constraint factor matrix is used to suppress background redundant information and enhance the significance of the fire point features.

[0066] Step 204: Add the first fusion feature to the first feature to obtain a target fusion feature.

[0067] Finally, directly combine the first fusion feature Add to the first feature , we can get the target fusion feature (corresponding to + ).

[0068] See also Figure 3 , the data flow is shown below. Perform Tucker decomposition to obtain the core tensor , the orthogonal first factor matrix And the second factor matrix At the same time, the first mapping data Input into the convolutional network to get the first feature . Second mapping data Input into the convolutional network to obtain the second feature Then, the second feature With core tensors , the first factor matrix And the second factor matrix fusion , get the first fusion feature Finally, the first fusion feature With the first feature Add them together to get the target fusion features.

[0069] In one embodiment, when the scales of the first fused feature and the first feature are not uniform, the method further includes: performing an edge-filling operation of zero values on the first fused feature and a feature with a smaller scale in the first feature.

[0070] That is, the edge-filling zero-value operation can make the fused features cover all spatial positions, avoiding the fire point positions being omitted due to size clipping.

[0071] Optionally, in one embodiment, the method further includes: if a fire point is detected, determining the target fire-fighting equipment closest to the fire point based on the positioning data of the fire point, and uploading the detection data to the control terminal.

[0072] The target fire-fighting equipment is the fire-fighting equipment closest to the fire point; and the detection data includes the approval control instructions for the target fire-fighting equipment and the positioning data of the fire point.

[0073] In the embodiments of the present application, the firefighting equipment can specifically be sprinkler heads, water guns, fire-fighting robots, etc. In the embodiments of the present application, by detecting the fire point in real time, automatically matching the nearest target firefighting equipment, and uploading the detection data to the control terminal to shorten the response control data, this solution can trigger the equipment to start within seconds, thereby suppressing the initial spread of the fire.

[0074] Optionally, the above convolutional network may include three convolutional layers.

[0075] In summary, the existing system cannot detect open flames when the diameter of the open flame is less than 10 cm and there is little smoke. The present invention uses the images collected by the network camera to detect the fire point to achieve early detection of fire. According to the characteristics of the fire point on the image, the present application uses Tucker decomposition to decompose the image background obtained by the pixel aggregation algorithm into core tensors. and two orthogonal factor matrices and , then the image background after three layers of convolution and the orthogonal components obtained by decomposition and the core tensor are fused, and then fused with the target image after three layers of convolution. Finally, the detection head is used to detect whether there is a fire point to implement corresponding processing measures.

[0076] Optionally, an embodiment of the present application also provides an auxiliary detection method for early fire conditions caused by smoke, that is, the above method may also include: obtaining a first image of the area to be detected; inputting the first image into the improved CSPNet model to obtain a first feature map; wherein the improved CSPNet model adopts a convolution module based on regional attention; and using a detection head to detect the first feature map to detect smoke.

[0077] Specifically, the embodiments of this application replace the conventional convolution in CSPNet with region-attention-based convolution (RAC). RAC uses an attention mechanism to accurately assess the importance of each feature point within the receptive field, thereby reducing information sparsity caused by parameter sharing. In other words, the RAC-enhanced CSPNet enhances the extraction of weak smoke features, thereby improving detection accuracy.

[0078] Alternatively, the composition of the improved CSPNet model can be found in Figure 4 , the following is a complete description of the process: the first image image of the area to be detected is sent to the improved CSPNet model, first A group convolution is applied to the first image image to extract the feature representation of each feature point within the receptive field to obtain the dimension where H, W, and C represent the height, width, and number of channels, respectively.

[0079] This process is followed by batch normalization and sigmoid activation function, and the shape is reshaped to get the size of The receptive field spatial characteristics.

[0080] Next, the spatial features extracted from the receptive field are used for spatial attention calculation. First, the maximum pooling and average pooling are used, and the image is used to obtain the size of The feature maps are then processed using convolution and sigmoid activation. The Squeeze and Excitation method is used as a channel attention mechanism, consisting of global average pooling, a fully connected layer, and a sigmoid activation function. Finally, the spatial and channel attention maps are combined (directly added) and fed into the detection head.

[0081] Optionally, see Figure 5 The early fire detection method provided in the embodiment of the present application may further include: steps 501 to 503.

[0082] Step 501: Obtain all cameras in the coverage area.

[0083] Each camera captures an area to be detected.

[0084] The coverage area can specifically refer to a specific monitoring scenario, such as a company, factory, or shopping mall. Multiple cameras need to be installed within the coverage area to monitor the area. For example, if the coverage area is a company, multiple cameras need to be installed, such as in the office, restroom, hallway, and staircase. In this example, the company corresponds to the coverage area, while the office, restroom, hallway, and staircase correspond to the detection areas.

[0085] Step 502: Construct a fire warning map based on all cameras in the coverage area.

[0086] In the fire warning map, each camera is regarded as a node, and an edge is established between two nodes in the connected area.

[0087] For example, camera A monitors the company gate and camera B monitors the front desk, and an edge is established between the two nodes of camera A and camera B.

[0088] Step 503: Predict the direction of fire spread based on the fire warning map.

[0089] It should be noted that by constructing a fire alarm map, early warning of fire spread can be achieved, and the direction and trend of fire spread can be known.

[0090] Optionally, in one embodiment, the edges in the fire warning graph are provided with weights; the weights of the edges are determined by the connectivity type between two nodes, the physical distance between the two nodes, and / or the flammability properties of the materials between the two nodes.

[0091] Examples of connectivity types between two nodes include: open space = 3 (high risk), regular door = 2 (medium risk), fire door = 1 (low risk), ventilation opening (lower risk).

[0092] The longer the physical distance between two nodes, the longer it takes for the fire to spread.

[0093] The material flammability properties between two nodes can include: paper material = 3, wood = 2, metal material = 0.5, and so on.

[0094] Among them, a calculation formula for the weight E can be:

[0095] E = physical distance / speed of spread × A + connection type × B + material flammability properties × C;

[0096] Among them, A, B and C represent the contribution ratios respectively.

[0097] The final fire warning graph can be expressed as G(N, E), where N represents a node and E represents an edge.

[0098] The time when each camera detects the fire is recorded. If node N1 detects the fire earlier than the adjacent node N2 and it is consistent with the direction of fire spread (obtained from the displacement of the detection frame), it is inferred that the fire spreads from 1 to 2.

[0099] Optionally, in one embodiment, the above steps predict the direction of fire spread based on the fire warning map, including: when a target node with a fire is detected, determining the direction of fire spread based on the weights of the edges between other nodes adjacent to the target node in the fire warning map.

[0100] That is, if there are multiple adjacent nodes, the one with the largest edge weight is selected as the potential area. As camera nodes gradually detect fires, the fire status in the topology map is dynamically updated. Upon fire detection, both the current fire location (the area monitored by the camera) and the fire spread (inferred from the map) are output simultaneously, allowing personnel to take timely countermeasures.

[0101] See also Figure 6 Based on the same inventive concept, the present embodiment further provides an early fire detection system 600, comprising:

[0102] The first acquisition module 601 is used to acquire an image group captured by a camera on the area to be detected and form a target tensor; wherein the image group includes multiple frames of images, and each frame of image includes an identified image background and a detection object.

[0103] The second acquisition module 602 is used to obtain first mapping data and second mapping data; wherein the first mapping data is the mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is the mapping of all detection objects in the image group in the target tensor.

[0104] The first processing module 603 is configured to perform Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix.

[0105] The second processing module 604 is used to input the first mapping data and the second mapping data into the convolutional network respectively, and fuse the output results with the core tensor, the first factor matrix and the second factor matrix to obtain target fusion features.

[0106] The detection module 605 is used to detect the target fusion features using a detection head to detect and locate the fire point.

[0107] See also Figure 7 Based on the same inventive concept, the present invention provides a module frame of an electronic device 700 that applies the above method. The electronic device 700 includes: at least one processor 701 ( Figure 7 Only one is shown), a memory 702, a computer program 703 stored in the memory 702 and executable on at least one processor 701, wherein the processor 701 implements the steps of the method in any of the foregoing embodiments when executing the computer program 703.

[0108] The electronic device 700 can be a server, a personal computer, a notebook computer, etc.

[0109] Those skilled in the art will understand that Figure 7 This is merely an example of the electronic device 700 and does not constitute a limitation on the electronic device 700 . The electronic device 700 may include more or fewer components than shown in the figure, or may combine certain components, or may include different components.

[0110] The processor 701 may be a central processing unit (CPU), or may be another general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor.

[0111] In some embodiments, the memory 702 may be an internal storage unit of the electronic device 700, such as a hard disk or memory of the electronic device 700. In other embodiments, the memory 702 may also be an external storage device of the electronic device 700, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. equipped on the electronic device 700. Furthermore, the memory 702 may include both an internal storage unit of the electronic device 700 and an external storage device.

[0112] It should be noted that the above-mentioned systems, equipment, etc. are based on the same concept as the method embodiments of this application. The modules designed for the systems, the steps performed by the equipment, and the technical effects brought about can all be found in the method embodiment section and will not be repeated here.

[0113] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0114] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0115] An embodiment of the present application provides a computer program product. When the computer program product is run on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned various method embodiments when executing the computer program product.

[0116] If the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to the camera / electronic device, a recording medium, computer memory, read-only memory (ROM), random access memory (RAM), an electrical carrier signal, a telecommunications signal, and a software distribution medium. Examples include a USB flash drive, a removable hard drive, a magnetic disk, or an optical disk.

[0117] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0118] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0119] In the embodiments provided in this application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely schematic. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0120] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0121] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. 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. 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 various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for detecting early-stage fire, characterized in that: include: Obtain an image group captured by a camera of the area to be detected and form a target tensor; wherein the image group includes multiple frames, and each frame includes the identified image background and the detection object; Acquire first mapping data and second mapping data; wherein the first mapping data is a mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is a mapping of all detection objects in the image group in the target tensor; Performing Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix; Inputting the first mapping data and the second mapping data into a convolutional network respectively, and fusing the output results with the core tensor, the first factor matrix, and the second factor matrix to obtain a target fused feature; Using a detection head to detect the target fusion features to detect and locate the fire point; Wherein, the first mapping data and the second mapping data are respectively input into the convolutional network, and the output result is fused with the core tensor, the first factor matrix and the second factor matrix to obtain the target fusion feature, including: inputting the first mapping data into the convolutional network to obtain the first feature; inputting the second mapping data into the convolutional network to obtain the second feature; wherein the output result includes the first feature and the second feature; fusing the second feature with the core tensor, the first factor matrix and the second factor matrix to obtain the first fusion feature; adding the first fusion feature to the first feature to obtain the target fusion feature.

2. The method for detecting early fire according to claim 1, characterized in that: When the scales of the first fused feature and the first feature are not consistent, the method further includes: Performing a zero-value edge filling operation on the first fused feature and a feature with a smaller scale among the first features.

3. The early fire detection method according to claim 1, characterized in that: The method further comprises: If a fire point is detected, the target fire-fighting equipment closest to the fire point is determined based on the positioning data of the fire point, and the detection data is uploaded to the control terminal; The target fire-fighting equipment is the fire-fighting equipment closest to the fire point; The detection data includes the approval control instructions for the target fire-fighting equipment and the location data of the fire point.

4. The method for detecting early fire according to claim 1, characterized in that: The convolutional network includes three convolutional layers.

5. The method for detecting early fire according to claim 1, characterized in that: The method further comprises: Acquiring a first image of the scene to be detected; Inputting the first image into an improved CSPNet model to obtain a first feature map; wherein the improved CSPNet model adopts a convolution module based on regional attention; The detection head is used to detect the first characteristic image to detect smoke.

6. The method for detecting early fire according to claim 1, characterized in that: The method further comprises: Get all cameras in the coverage area; each camera captures an area to be detected; Constructing a fire warning map based on all cameras in the coverage area; wherein each camera in the fire warning map is regarded as a node, and an edge is established between two nodes connected in the area; Based on the fire warning map, the direction of fire spread is predicted.

7. The method for detecting early-stage fire according to claim 6, characterized in that: The edges in the fire warning graph are provided with weights; The weight of an edge is determined by the connectivity type between two nodes, the physical distance between two nodes, and / or the flammability properties of the materials between two nodes.

8. The method for detecting early-stage fire according to claim 7, characterized in that: The predicting of the fire spread direction based on the fire warning map includes: When a target node with a fire is detected, the direction of the fire spread is determined based on the weights of the edges between other nodes adjacent to the target node in the fire warning map.

9. An early fire detection system, characterized in that: include: A first acquisition module is configured to acquire an image group captured by a camera of the area to be detected and form a target tensor; wherein the image group includes multiple frames of images, and each frame of the image includes an identified image background and a detection object; A second acquisition module is configured to acquire first mapping data and second mapping data; wherein the first mapping data is a mapping of all image backgrounds in the image group in the target tensor, and the second mapping data is a mapping of all detected objects in the image group in the target tensor; a first processing module, configured to perform Tucker decomposition on the first mapping data to obtain a core tensor, an orthogonal first factor matrix, and a second factor matrix; a second processing module, configured to input the first mapping data and the second mapping data into a convolutional network respectively, and fuse the output results with the core tensor, the first factor matrix, and the second factor matrix to obtain a target fusion feature; A detection module, configured to detect the target fusion features using a detection head to detect and locate the fire point; The second processing module is further used to input the first mapping data into the convolutional network to obtain a first feature; input the second mapping data into the convolutional network to obtain a second feature; wherein the output result includes the first feature and the second feature; fuse the second feature with the core tensor, the first factor matrix and the second factor matrix to obtain a first fused feature; add the first fused feature to the first feature to obtain the target fused feature.

Images