CN112215122B - Fire detection method, system, terminal and storage medium based on video image target detection - Google Patents

Abstract

The application relates to a fire detection method, a fire detection system, a fire detection terminal and a fire detection storage medium based on video image target detection. Comprising the following steps: converting an original natural image into a dust haze image and a sand dust image by adopting a data enhancement algorithm based on an atmospheric scattering model, and generating a data set for training the model; constructing a convolutional neural network model LFNT, inputting a data set into the LFNT model for iterative training to obtain optimal model parameters; the convolutional neural network model LFNT comprises a skeleton feature extraction model, a main feature extraction model and a variable-scale feature fusion model; the skeleton feature extraction model extracts main features of an input image through convolution of three different scales; the main feature extraction model is used for carrying out further feature extraction on main features to generate three groups of feature graphs; and the variable scale feature fusion model carries out self-adaptive fusion on the three groups of feature graphs, and outputs a detection result. The method can improve the robustness of the model in abnormal weather such as sand dust, dust haze and the like, and enable the model to obtain a better detection result.

Description

Fire detection method, system, terminal and storage based on video image target detection medium

Technical field

This application belongs to the field of fire detection technology, and particularly relates to a fire detection method, system, terminal and storage medium based on video image target detection.

Background technique

Fire detection plays a vital role in safety monitoring. At present, the traditional fire detection method is based on the image prior method. This method detects fire based on the color and shape of the image. However, the robustness and bit error rate of color and motion features are often limited by preset parameters. The impact makes it impossible to apply in complex environments, and positioning accuracy is easily affected by regions.

Monitoring is a tedious and time-consuming task, especially in an uncertain monitoring environment, which has great uncertainty in time, space and even scale. Sensor-based detectors have limited performance in terms of bit error rate and sensing range; therefore, they are unable to detect distant or small fires. In recent years, with the rapid development of deep learning technology, convolutional neural networks (CNN) have been applied to fire detection. However, existing fire detection methods based on deep learning still have the following shortcomings:

1. Methods based on deep learning require a large amount of remote sensing images as training data. Due to the scarcity of real remote sensing images, model training is very challenging.

2. The fire detection model based on deep learning is too large and is not suitable for equipment with limited resources.

3. The complexity of the existing algorithms is too high and cannot be detected in real time.

4. The anti-interference ability is weak and is easily affected by harsh monitoring environments such as haze and dust.

5. Most fire detection algorithms only focus on a single environment, therefore, higher error rates will occur in uncertain environments.

In summary, existing fire detection methods have great room for improvement in terms of algorithm complexity, application scenario range, and model size.

Contents of the invention

This application provides a fire detection method, system, terminal and storage medium based on video image target detection, aiming to solve one of the above technical problems in the prior art at least to a certain extent.

In order to solve the above problems, this application provides the following technical solutions:

A fire detection method based on video image target detection, including:

A data enhancement algorithm based on the atmospheric scattering model is used to convert the original natural images into haze images and dust images to generate a data set for training the model;

Construct a convolutional neural network model LFNet, input the data set into the LFNet model for iterative training, and obtain the optimal model parameters; the skeleton feature extraction models respectively adopt convolution extraction at 3*3, 5*5 and 7*7 scales. Input the features of the image and obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively; the main feature extraction model further extracts the main features and generates sizes of 52*52, 26*26 and 13 respectively. *13 three sets of feature maps; the variable scale feature fusion model maps the three sets of feature maps to different convolution kernels and step sizes for convolution, and splices all convolutions of the same size to obtain three sets of feature maps, using channel-based The attention mechanism operates the three sets of feature maps to obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively, which are used to detect small, medium and large objects respectively; the data set is input into the LFNet model for Iterative training also includes: selecting mean square error and cross entropy as loss functions respectively for model optimization;

The convolutional neural network model LFNet includes a skeleton feature extraction model, a main feature extraction model and a variable scale feature fusion model; the skeleton feature extraction model extracts the main features of the input image through convolution at three different scales; the main features The extraction model is used to further extract the main features and generate three groups of feature maps; the variable scale feature fusion model adaptively fuses the three groups of feature maps and outputs detection results;

Input the fire image to be detected into the trained LFNet model, and output the fire location area and fire type of the fire image to be detected through the LFNet model.

The technical solutions adopted by the embodiments of the present application also include: before using the data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a haze image and a dust image, the following steps are included:

Obtain original natural images; the original natural images include non-alarm images without fire alarm areas and real fire alarm images.

The technical solutions adopted by the embodiments of this application also include: The use of a data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a haze image includes:

The atmospheric scattering model uses at least two transmission rates to simulate and generate haze images of different concentrations; the haze image imaging formula is:

I(x)＝J(x)t(x)+ɑ(1-t(x))

In the above formula, I(x) is the simulated haze image, J(x) is the input haze-free image, ɑ is the atmospheric light value, and t(x) is the scene transmission rate.

The technical solutions adopted by the embodiments of this application also include: The use of a data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a sand and dust image includes:

The atmospheric scattering model uses a fixed transmittance and atmospheric light value, and combines three colors to simulate and generate sand and dust images of different concentrations; the sand and dust image simulation formula is:

D(x)=J(x)t(x)+a(C(x)*(1-t(x)))

In the above formula, D(x) is the simulated dust image, J(x) is the input haze-free image, and C(x) is the color value.

The technical solutions adopted in the embodiments of this application also include: the loss function is specifically:

The brightness, dark channel value and R channel data of the path in the fire area are counted. The statistical data are regarded as a combustion histogram prior and written as the formula of CHP:

In the above formula, R(x) represents the R channel of the image, SCP(x) is the difference between the brightness and dark channels of the image, w is the width of the histogram, and h is the height of the histogram;

SCP(x)=||v(x)-DCP(x)||

In the above formula, v(x) is the brightness of the image, and DCP(x) is the value of the dark channel of the image;

L _CHP =||CHP(I)-CHP(R)|| ²

In the above formula, CHP represents the combustion histogram prior, CHP(I) and CHP(R) represent the CHP value of the area selected by the target detection algorithm and the marked area respectively;

The loss function is a weighted sum of three different loss functions:

L _CHP =βL _CE +γL _MSE +δL _CHP

In the above formula, L _CHP is the final loss function, _ICE is the cross entropy loss function, L _MSE is the mean square error loss function, and L _CHP is the combustion histogram prior loss.

Another technical solution adopted by the embodiment of the present application is: a fire detection system based on video image target detection, including:

Dataset building module: used to convert original natural images into haze images and dust images using a data enhancement algorithm based on the atmospheric scattering model, and generate a data set for training the model;

LFNet model training module: used to construct the convolutional neural network model LFNet, input the data set into the LFNet model for iterative training, and obtain the optimal model parameters; the skeleton feature extraction models adopt 3*3, 5*5 and 7 respectively. *7-scale convolution extracts the features of the input image and obtains feature maps with sizes of 13*13, 26*26 and 52*52 respectively; the main feature extraction model further extracts the main features and generates features with sizes of 52*. Three sets of feature maps: 52, 26*26, and 13*13; the variable scale feature fusion model maps the three sets of feature maps to different convolution kernels and step sizes for convolution, and splices all convolutions of the same size to obtain three sets of feature maps. A group of feature maps uses a channel-based attention mechanism to operate the three groups of feature maps to obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively, which are used to detect small, medium and large objects respectively; Inputting the data set into the LFNet model for iterative training also includes: selecting the mean square error and cross entropy as the loss function respectively for model optimization;

The convolutional neural network model LFNet includes a skeleton feature extraction model, a main feature extraction model and a variable scale feature fusion model; the skeleton feature extraction model extracts the main features of the input image through convolution at three different scales; the main features The extraction model is used to further extract the main features and generate three sets of feature maps; the variable scale feature fusion model adaptively fuses the three sets of feature maps and outputs detection results; the detection results include fire The fire location area of the image and the type of fire.

Another technical solution adopted by the embodiment of the present application is: a terminal, the terminal includes a processor and a memory coupled to the processor, wherein,

The memory stores program instructions for implementing the fire detection method based on video image target detection;

The processor is configured to execute the program instructions stored in the memory to control fire detection based on video image target detection.

Another technical solution adopted by the embodiment of the present application is: a storage medium that stores program instructions executable by a processor, and the program instructions are used to execute the fire detection method based on video image target detection.

Compared with the existing technology, the beneficial effects produced by the embodiments of the present application are that: the fire detection methods, systems, terminals and storage media based on video image target detection of the embodiments of the present application use a data enhancement algorithm based on the atmospheric scattering model to transform the original image into Convert to images affected by haze or sand to varying degrees, generate a data set for training the model, and build a convolutional neural network model LFNet suitable for fire smoke detection in uncertain environments, which can improve the model's performance in sand, dust and haze The robustness under abnormal weather conditions enables the model to obtain better detection results. At the same time, due to the small size of the LFNet model in the embodiment of the present application, the calculation cost can be reduced, and it is beneficial for the LFNet model to be applied to devices with limited resources.

Description of the drawings

Figure 1 is a flow chart of a fire detection method based on video image target detection according to an embodiment of the present application;

Figure 2 is a schematic diagram of the haze and dust image simulation effects based on the atmospheric scattering model according to the embodiment of the present application;

Figure 3 is a framework diagram of the convolutional neural network model according to the embodiment of the present application;

Figure 4 is a structural diagram of the variable scale feature fusion model according to the embodiment of the present application;

Figure 5 is a structural diagram of the channel-based attention mechanism according to the embodiment of the present application;

Figure 6 is a schematic structural diagram of a fire detection system based on video image target detection according to an embodiment of the present application;

Figure 7 is a schematic structural diagram of a terminal according to an embodiment of the present application;

Figure 8 is a schematic structural diagram of a storage medium according to an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

Please refer to Figure 1, which is a flow chart of a fire detection method based on video image target detection according to an embodiment of the present application. The fire detection method based on video image target detection in the embodiment of the present application includes the following steps:

S10: Obtain original natural images;

In this step, the original natural images obtained include 293 non-alarm images without fire alarm areas and 5073 real fire alarm images. Using non-alarm images can improve the robustness of the training algorithm to non-alarm targets and reduce the bit error rate of the detector. Utilizing real fire alarm images can improve the detection capabilities of target detection models.

S20: Use a data enhancement algorithm based on the atmospheric scattering model to convert original natural images into new synthetic images affected by different types and degrees of abnormal weather, and generate a data set for training the model;

In this step, because existing intelligent monitoring algorithms usually ignore the impact of abnormal weather such as haze or sand dust on performance, the monitoring algorithm has poor robustness under uncertain climate conditions. In order to solve the above shortcomings, the present invention considers the impact of abnormal weather on fire detection algorithms, and simulates haze images and dust images of different degrees through a data enhancement method based on the atmospheric scattering model, thereby converting original natural images into images affected by different Based on new synthetic images affected by haze or dust weather, a large-scale benchmark data set is constructed for training and testing fire detection models to improve the robustness of the target detection model under abnormal weather such as sand, dust and haze.

Further, please refer to Figure 2, which is a schematic diagram of the haze and dust image simulation effects based on the atmospheric scattering model according to the embodiment of the present application, in which (a) is the original image, (b), (c) and (d) are respectively Haze images synthesized by atmospheric scattering models with different transmission rates. (e), (f) and (g) are sand and dust images simulated using fixed transmittance and atmospheric light values, combined with three different colors respectively. The haze image imaging formula is:

I(x)＝J(x)t(x)+ɑ(1-t(x)) (1)

In formula (1), I(x) is the simulated haze image, J(x) is the input haze-free image, ɑ is the atmospheric light value, t(x) is the scene transmission rate, which describes the scene transmission rate in the view. The part that is not scattered and reaches the camera sensor. In order to simulate haze weather with different concentrations, the embodiment of the present application sets the atmospheric light value ɑ to 0.8, and sets the transmittance to 0.8, 0.6 and 0.4 respectively.

Since depth information does not play a major role in image dusting tasks, the transfer is assumed not to change with the depth of the image. Through a priori statistics, the embodiment of this application selects three colors suitable for simulating sand and dust images for simulation respectively. The sand and dust image simulation formula is:

D(x)=J(x)t(x)+a(C(x)*(1-t(x))) (2)

In formula (2), D(x) is the simulated dust image, J(x) is the input haze-free image, and C(x) is the selected color value.

S30: Construct the convolutional neural network model LFNet;

In the embodiment of this application, the framework of the convolutional neural network model is shown in Figure 3. LFNet is composed of common convolutional layers, bottleneck building blocks, parameter correction linear units, group normalization, etc., including: skeleton feature extraction model, main feature extraction model and variable scale feature fusion model. The specific functions of each model are:

Skeleton feature extraction model: used to extract the main features of the input image. In order to extract richer image features, we first use 3*3, 5*5 and 7*7 scale convolutions to extract the features of the input image, expand the acceptance field, and extract more image features. After convolution at three different scales, feature maps with sizes of 13*13, 26*26 and 52*52 are obtained. Based on the above, by using multi-scale convolution for feature map extraction, feature information of different sizes around the pixels can be extracted, which is particularly important for fire images.

Main feature extraction model: used to further extract features from the main features extracted by the skeleton feature extraction model, and generate three sets of feature maps with sizes of 52*52, 26*26, and 13*13. Each small-sized feature The maps are extracted from the larger-sized feature maps of the upper layer, and each convolution block is extracted by a layer of convolution structure and a five-layer residual structure.

Variable scale feature fusion model: used to use variable scale feature fusion (VSFF) to concatenate the features extracted by the main feature extraction model, and then use convolution to extract features and perform adaptive fusion of the features. The structure of the variable-scale feature fusion model is shown in Figure 4. In order to fuse the feature maps extracted by convolution of different scales, three sets of feature map maps are fused, and the functions of 13*13 and 26*26 are extended to 52*52. The three inputs are feature maps with sizes of 13*13, 26*26, and 52*52 respectively. The three feature maps of different sizes are mapped to different convolution kernels and step sizes for convolution, allowing upsampling or downsampling. Become two other sizes. Finally, all convolutions of the same size are concatenated to obtain three sets of feature maps. Since the feature map obtained by splicing contains richer image features, the model positioning can be made more accurate.

Furthermore, the embodiment of the present application uses a channel-based attention mechanism to operate the three sets of feature maps extracted in VSFF. The channel-based attention mechanism can be viewed as a process of weighting feature maps according to their importance. For example, in a set of 24*13*13 convolutions, the channel-based attention mechanism will determine which feature map in the set has a more significant impact on the prediction result, and then increase the weight of that part. With the help of the attention mechanism, three fusions are performed to obtain feature maps with sizes of 13*13, 26*26 and 52*52, which are used to detect small, medium and large objects respectively. The detailed structure of the channel-based attention mechanism is shown in Figure 5.

Based on the above structure, the size of the LFNet model in the embodiment of the present application is very small (22.5M), but it occupies a leading position in both quantitative and qualitative evaluation, reducing the computational cost and conducive to the application of LNet to resource-limited devices.

S40: Input the data set into the LFNet model for iterative training to obtain optimal model parameters;

In this step, during the model training process, the LFNet model has two tasks: one is to accurately locate the alarm area in the image; the other is to classify the disaster type in the alarm area. In order to enable the model to better complete these two tasks, the embodiment of this application selects mean square error (MSE) and cross entropy (CE) as the loss function to guide network optimization. This loss function is based on a large number of statistics on different fire images or videos. , can help LFNet effectively detect fire areas.

Specifically, after conducting a large number of experiments on various fire images, it was found that in the smoke area, the absolute value of the difference between the brightness and dark channel values is higher than that in other areas, and the R channel in the fire area is higher than that in the non-fire area, that is, the brightness of the path, The dark channel value and R channel vary with the fire danger area. The smoke concentration increases with the absolute value of the difference between the brightness and the dark channel. The visual characteristics of the fire are closely related to the pixel value of the R channel. Based on the above characteristics, the embodiment of the present application regards these statistical data as combustion histogram prior (CHP), and based on these statistical data, it is written as the formula of CHP:

In formula (3), R(x) represents the R channel of the image, SCP(x) is the difference between the brightness and dark channels of the image, w is the width of the histogram, and h is the height of the histogram; it can also be written as :

SCP(x)＝||v(x)-DCP(x)|| (4)

In formula (4), v(x) is the brightness of the image, and DCP(x) refers to the value of the dark channel of the image.

L _CHP =||CHP(I)-CHP(R)|| ² (5)

In formula (5), CHP represents the combustion histogram prior, and CHP(I) and CHP(R) represent the CHP values of the area selected by the target detection algorithm and the area marked in the ground truth respectively.

The final loss function is the weighted sum of three different loss functions: cross entropy loss function, mean square error loss function and combustion histogram prior loss function. The formula is:

L _CHP =βL _CE +γL _MSE +δL _CHP (6)

In formula (6), L _CHP is the final loss function, L _CE is the cross entropy loss function, L _MSE is the mean square error loss function, L _CHP is the combustion histogram prior loss, β, γ and δ are set to 0.25 respectively. , 0.25 and 0.5.

S50: Input the fire image to be detected into the trained LFNet model, and output the fire location area and fire type of the fire image to be detected through the LFNet model.

Please refer to FIG. 6 , which is a schematic structural diagram of a fire detection system based on video image target detection according to an embodiment of the present application. The fire detection system 40 based on video image target detection in the embodiment of the present application includes:

Data set construction module 41: used to convert original natural images into haze images and dust images using a data enhancement algorithm based on the atmospheric scattering model, and generate a data set for training the model;

LFNet model training module 42: used to construct the convolutional neural network model LFNet, input the data set into the LFNet model for iterative training, and obtain optimal model parameters; the skeleton feature extraction models respectively adopt 3*3, 5*5 and The 7*7 scale convolution extracts the features of the input image and obtains feature maps with sizes of 13*13, 26*26 and 52*52 respectively; the main feature extraction model further extracts the main features and generates sizes of 52 Three sets of feature maps of *52, 26*26, and 13*13; the variable scale feature fusion model maps the three sets of feature maps to different convolution kernels and step sizes for convolution, and splices all convolutions of the same size to obtain Three sets of feature maps are used to operate the three sets of feature maps using a channel-based attention mechanism to obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively, which are used to detect small, medium and large objects respectively; so The above-mentioned input of the data set into the LFNet model for iterative training also includes: selecting the mean square error and cross entropy as the loss function for model optimization;

The convolutional neural network model LFNet includes a skeleton feature extraction model, a main feature extraction model and a variable scale feature fusion model; the skeleton feature extraction model extracts the main features of the input image through convolution at three different scales; the main features The extraction model is used to further extract the main features and generate three groups of feature maps; the variable scale feature fusion model adaptively fuses the three groups of feature maps and outputs detection results;

Model optimization module 43: used to select mean square error and cross entropy as loss functions respectively for model optimization.

Please refer to Figure 7, which is a schematic structural diagram of a terminal according to an embodiment of the present application. The terminal 50 includes a processor 51 and a memory 52 coupled to the processor 51 .

The memory 52 stores program instructions for implementing the above fire detection method based on video image target detection.

The processor 51 is used to execute program instructions stored in the memory 52 to control fire detection based on video image target detection.

The processor 51 may also be called a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. . A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

Please refer to FIG. 8 , which is a schematic structural diagram of a storage medium according to an embodiment of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods. The program file 61 can be stored in the above storage medium in the form of a software product and includes a number of instructions to make a computer device (can It is a personal computer, server, or network device, etc.) or processor that executes all or part of the steps of the various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. , or terminal equipment such as computers, servers, mobile phones, tablets, etc.

The fire detection method, system, terminal and storage medium based on video image target detection in the embodiment of the present application convert the original image into an image affected by haze or sand to varying degrees by using a data enhancement algorithm based on the atmospheric scattering model to generate a The data set for training the model and constructing the convolutional neural network model LFNet suitable for fire smoke detection in uncertain environments can improve the robustness of the model in abnormal weather such as sand, dust and haze, and enable the model to obtain better detection result. At the same time, due to the small size of the LFNet model in the embodiment of the present application, the calculation cost can be reduced, and it is beneficial for the LFNet model to be applied to devices with limited resources.

The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined in this application may be practiced in other embodiments without departing from the spirit or scope of the application. Therefore, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (8)

1. A fire detection method based on video image target detection, characterized by including: A data enhancement algorithm based on the atmospheric scattering model is used to convert the original natural images into haze images and dust images to generate a data set for training the model; Construct a convolutional neural network model LFNet, input the data set into the LFNet model for iterative training, and obtain the optimal model parameters; the skeleton feature extraction models respectively adopt convolution extraction at 3*3, 5*5 and 7*7 scales. Input the features of the image and obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively; the main feature extraction model further extracts the main features and generates sizes of 52*52, 26*26 and 13 respectively. *13 three sets of feature maps; the variable scale feature fusion model maps the three sets of feature maps to different convolution kernels and step sizes for convolution, and splices all convolutions of the same size to obtain three sets of feature maps, using channel-based The attention mechanism operates the three sets of feature maps to obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively, which are used to detect small, medium and large objects respectively; the data set is input into the LFNet model for Iterative training also includes: selecting mean square error and cross entropy as loss functions respectively for model optimization; The convolutional neural network model LFNet includes a skeleton feature extraction model, a main feature extraction model and a variable scale feature fusion model; the skeleton feature extraction model extracts the main features of the input image through convolution at three different scales; the main features The extraction model is used to further extract the main features and generate three groups of feature maps; the variable scale feature fusion model adaptively fuses the three groups of feature maps and outputs detection results; Input the fire image to be detected into the trained LFNet model, and output the fire location area and fire type of the fire image to be detected through the LFNet model. 2. The fire detection method based on video image target detection according to claim 1, characterized in that the use of a data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a haze image and a dust image includes: Obtain original natural images; the original natural images include non-alarm images without fire alarm areas and real fire alarm images. 3. The fire detection method based on video image target detection according to claim 1 or 2, characterized in that the use of a data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a haze image includes: The atmospheric scattering model uses at least two transmission rates to simulate and generate haze images of different concentrations; the haze image imaging formula is: I(x)＝J(x)t(x)+ɑ(1-t(x)) In the above formula, I(x) is the simulated haze image, J(x) is the input haze-free image, ɑ is the atmospheric light value, and t(x) is the scene transmission rate. 4. The fire detection method based on video image target detection according to claim 3, characterized in that the use of a data enhancement algorithm based on the atmospheric scattering model to convert the original natural image into a sand and dust image includes: The atmospheric scattering model uses a fixed transmittance and atmospheric light value, and combines three colors to simulate and generate sand and dust images of different concentrations; the sand and dust image simulation formula is: D(x)=J(x)t(x)+a(C(x)*(1-t(x))) In the above formula, D(x) is the simulated dust image, J(x) is the input haze-free image, and C(x) is the color value. 5. The fire detection method based on video image target detection according to claim 1, characterized in that the loss function is specifically: The brightness, dark channel value and R channel data of the path in the fire area are counted. The statistical data are regarded as a combustion histogram prior and written as the formula of CHP: In the above formula, R(x) represents the R channel of the image, SCP(x) is the difference between the brightness and dark channels of the image, w is the width of the histogram, and h is the height of the histogram; SCP(x)=||v(x)-DCP(x)|| In the above formula, v(x) is the brightness of the image, and DCP(x) is the value of the dark channel of the image; L _CHP =||CHP(I)-CHP(R)|| ² In the above formula, CHP represents the combustion histogram prior, CHP(I) and CHP(R) represent the CHP value of the area selected by the target detection algorithm and the marked area respectively; The loss function is a weighted sum of three different loss functions: L _CHP =βL _CE +γL _MSE +δL _CHP In the above formula, L _CHP is the final loss function, L _CE is the cross entropy loss function, L _MSE is the mean square error loss function, and L _CHP is the combustion histogram prior loss. 6. A fire detection system based on video image target detection, characterized by including: Dataset building module: used to convert original natural images into haze images and dust images using a data enhancement algorithm based on the atmospheric scattering model, and generate a data set for training the model; LFNet model training module: used to construct the convolutional neural network model LFNet, input the data set into the LFNet model for iterative training, and obtain the optimal model parameters; the skeleton feature extraction models adopt 3*3, 5*5 and 7 respectively. *7-scale convolution extracts the features of the input image and obtains feature maps with sizes of 13*13, 26*26 and 52*52 respectively; the main feature extraction model further extracts the main features and generates features with sizes of 52*. Three sets of feature maps: 52, 26*26, and 13*13; the variable scale feature fusion model maps the three sets of feature maps to different convolution kernels and step sizes for convolution, and splices all convolutions of the same size to obtain three sets of feature maps. A group of feature maps uses a channel-based attention mechanism to operate the three groups of feature maps to obtain feature maps with sizes of 13*13, 26*26 and 52*52 respectively, which are used to detect small, medium and large objects respectively; Inputting the data set into the LFNet model for iterative training also includes: selecting the mean square error and cross entropy as the loss function respectively for model optimization; The convolutional neural network model LFNet includes a skeleton feature extraction model, a main feature extraction model and a variable scale feature fusion model; the skeleton feature extraction model extracts the main features of the input image through convolution at three different scales; the main features The extraction model is used to further extract the main features and generate three sets of feature maps; the variable scale feature fusion model adaptively fuses the three sets of feature maps and outputs detection results; the detection results include fire The fire location area of the image and the type of fire. 7. A terminal, characterized in that the terminal includes a processor and a memory coupled to the processor, wherein, The memory stores program instructions for implementing the fire detection method based on video image target detection according to any one of claims 1 to 5; The processor is configured to execute the program instructions stored in the memory to control fire detection based on video image target detection. 8. A storage medium, characterized in that it stores program instructions executable by a processor, and the program instructions are used to execute the fire detection method based on video image target detection according to any one of claims 1 to 5.