Gas leakage detection visual identification method based on motion feature fusion

Abstract

The invention discloses a gas leakage detection visual identification method based on motion feature fusion. The method comprises the following steps: S1, acquiring infrared gas leakage image sample data; s2, performing frame extraction and positive and negative sample labeling on the infrared gas leakage image sample data; s3, preprocessing the image, and extracting motion characteristics of gas leakage of the image; s4, fusing the labeled data and the motion features, and generating a pre-training weight model in advance by using a large amount of pre-collected existing infrared gas leakage data sets based on a yov5 infrared image detection method; on the basis of the pre-training model, Few-Shot learning of the current scene is carried out, and a fine tuning model of infrared gas leakage detection is obtained; and S5, applying the trained fine tuning model to an infrared gas leakage video reasoning test, and outputting a gas leakage prediction result.

Description

Technical Field

[0001] The invention relates to the technical field of fire detection and image-type fire detectors, and in particular to a gas leakage detection visualization recognition method based on motion feature fusion. Background Art

[0002] Traditional gas detection requires personnel to conduct on-site inspections with handheld tools at close range. Some places are not convenient for personnel to approach due to narrow areas or dangerous factors, which brings certain difficulties to gas detection. Compared with traditional gas detection methods, infrared thermal imaging gas detection (OGI) has a large detection range, long distance, high efficiency, and is intuitive and easy to understand. It can perform detection operations hundreds of meters away in dangerous or difficult-to-reach areas, and it is also more secure for personnel safety.

[0003] Infrared thermal imaging gas detection technology aims to quickly detect the presence of leaked gas and accurately locate the source of the leak. It is based on infrared movement, thermal imaging module or thermal imager, and selectively designs narrow-band window according to the characteristics of gas imaging. It only needs to detect and image one or several infrared bands covering the characteristic absorption peaks of the gas, and through special image filtering and enhancement processing, it can timely and intuitively display the gas leakage in the output video image, helping detection personnel to find the source of the leak, evaluate the diffusion trend of the gas and the possible harm. This type of infrared detection equipment is usually divided into two categories: refrigerated type and non-refrigerated type. Non-refrigerated infrared gas temperature detection equipment is widely used in many fields due to its simple structure, low cost and high reliability. It can be used in electricity, oil, natural gas, petrochemical, chemical, energy production, environmental monitoring and other industries.

[0004] Current product research results include but are not limited to motion extraction calculations on continuous video frames collected by infrared gas temperature detection equipment based on traditional machine learning methods such as frame difference, optical flow, and background subtraction to obtain the pixel area where the leaking gas is located; detection methods based on deep learning, by using CNN (convolutional neural network) to extract gas features in infrared images and train visual detection algorithm models to predict gas leak detection.

[0005] However, the inventors of this application have found that the above technology has at least the following technical problems:

[0006] First of all, the methods used by existing products based on traditional machine learning are mainly frame difference method and optical flow method.

[0007] The frame difference method detects the leakage area by comparing the difference between two or more consecutive frames. This method has low sensitivity and can only detect larger gas leaks, because smaller gas leaks may not cause enough difference between adjacent frames. At the same time, this method is easily affected by environmental factors, such as lighting changes, object movement, etc., which may cause large differences between adjacent frames, resulting in false alarms.

[0008] Optical flow detects leaks by analyzing the pattern of object motion in an image sequence. This method requires processing the entire video sequence, which has high computational complexity and may result in insufficient real-time performance. The performance of optical flow may be limited for non-rigid objects or scenes with large changes in lighting. If the motion of a gas leak is not obvious or is similar to background motion, optical flow may have difficulty accurately detecting it.

[0009] Secondly, existing deep learning-based detection methods require a large amount of high-quality training data. However, there is relatively little open-source infrared gas leakage data, and data collection is also very difficult, which makes it difficult to obtain sufficient sample data and train a high-quality model. In addition, the contrast of infrared images is relatively low, and the edges of leaking gas are blurred, making it difficult to distinguish from the background; the quality of infrared images is poor and will be affected by temperature fluctuations, lighting changes, similar background objects, etc., making it difficult to effectively identify leaking gas.

[0010] Finally, most of the existing products only perform some customized detection processing for a single specified scene, which is not suitable for scenes with large changes in the background environment or scenes with unclear movement of leaking gas; some only use deep learning methods, which are not suitable for scenes with small amounts of training data or scenes with difficulty in distinguishing between targets and backgrounds. These algorithms do not have the ability to generalize to multiple scenes, and a large amount of data must be collected for each scene, which increases the workload.

[0011] Therefore, there is currently a lack of a complete set of mature, efficient, highly feasible, and holistic solutions that can be actually applied to most social production and life, as well as how the solution can truly achieve universal and effective detection of infrared gas leaks in complex scenarios and give timely warnings to ensure safety. Based on the above technical difficulties, the present invention proposes a solution that can achieve both rapid real-time detection and high-precision detection. Summary of the invention

[0012] In response to the above technical problems, the present application provides a method for visual identification of gas leak detection based on motion feature fusion. By combining deep learning and traditional machine learning methods, the present invention only needs to collect a small amount of current scene data for Few-Shot learning in the context of infrared gas leakage data being difficult to collect, and can quickly train to obtain a detection model adapted to the current scene, greatly reducing the difficulty of data collection and model training. The obtained detection model has very high accuracy on the basis of meeting the requirements of fast real-time detection, and is suitable for most industrial application scenarios.

[0013] The present application provides a gas leak detection visualization recognition method based on motion feature fusion, which is characterized by comprising:

[0014] Step S1: Collect infrared gas leakage image sample data;

[0015] Step S2: extracting frames and labeling positive and negative samples of the infrared gas leakage image sample data;

[0016] Step S3: image preprocessing, extracting motion features of gas leakage in the image;

[0017] Step S4: Fusing the annotated data and motion features, based on the YOLOv5 infrared image detection method, a large number of pre-collected infrared gas leakage data sets are used to train and generate a pre-trained weight model; based on the pre-trained weight model, Few-Shot learning of the current scene is performed to obtain a fine-tuning model for infrared gas leakage detection;

[0018] Step S5: Use the trained fine-tuning model for infrared gas leakage video inference test and output the gas leakage prediction result.

[0019] Preferably, in step S2, the frame extraction and labeling of the infrared gas leakage image sample data includes the following steps:

[0020] S21: performing frame extraction processing on all infrared gas leakage videos collected by the infrared camera to obtain infrared gas leakage images;

[0021] S22: Use the labelImg tool to draw and annotate all infrared gas leakage images with rectangular boxes of leaking gas and easily confused background to obtain positive and negative sample annotation data.

[0022] Preferably, step S3 includes:

[0023] S31: Use opencv image processing technology and bilateral filtering to perform denoising, and then use CLAHE image enhancement technology to improve the image contrast;

[0024] S32: Select the first T frames of the video sequence and construct a background model based on the K-Nearest Neighbors algorithm. The method for constructing the background model is as follows:

[0025] (1)

[0026] In the formula, To estimate the density, represents the sample set, , represents background pixels, represents the foreground pixel, Indicates the number of sequential frames obtained from the video forward. The radius is The volume of the inner core of the hypersphere, is a kernel function, if ,but , otherwise 0, is the diameter of the hypersphere, Represents the nuclear volume The number of samples contained in the hypersphere, the diameter The size of is related to the sample size. The diameter of the hypersphere is gradually expanded with the current pixel sample as the center. until it contains samples, fixed nearest neighbors;

[0027] S33: At time t, the background model can determine whether each new pixel sample x belongs to the background or the foreground. The evaluation process is as follows:

[0028] (2)

[0029] in, To estimate the density, and and Directly proportional relationship; Approximately:

[0030] (3)

[0031] in, is the background model weight vector, at time m, using represents the foreground of the sample, Indicates the background of the sample;

[0032] S34: Update and ; Update by adding the current time sample and eliminating the oldest time sample , by judging Whether it belongs to the background, to update B;

[0033] S35: Obtain a foreground image after removing the background through the foreground detection algorithm based on K-Nearest Neighbors, and then use an image opening operation to remove noise, thereby obtaining motion feature information.

[0034] Preferably, the step S4 comprises:

[0035] S41: The convolution features extracted from the foreground motion image and the infrared image are fused into a two-layer network structure consisting of convolution Conv, batch normalization BN and activation function SiLU by adding, and then the yolov5 network model structure is spliced ​​to form a new network model structure that can simultaneously receive two parameters of the infrared image and the foreground motion image;

[0036] S42: several infrared gas leakage data sets collected in advance are processed in step S2 to obtain foreground motion images, and processed in step S3 to obtain labeled images, and are simultaneously sent to the new network model structure described in step S41 to train and generate a universal pre-trained weight model;

[0037] S43: A small amount of sample data collected for the current application scenario is subjected to the same data processing and data fusion method as the pre-trained data in step S42, and then sent to the new network model structure described in step S41 to perform Few-Shot fine-tuning training of the model to obtain a fine-tuned detection model suitable for the current scenario.

[0038] Preferably, in step S5, the model performing reasoning test comprises the following steps:

[0039] S51: Input the infrared gas leakage video, perform frame processing, perform background subtraction modeling, and extract motion feature information;

[0040] S52: Use the new network model structure obtained in S41 as the inference network structure, and the fine-tuned detection model obtained in S43 as the inference model, read the input frame data and motion feature information, infer the gas leakage position rectangle and confidence in the current frame image, compare with the preset threshold, and determine whether infrared gas leakage occurs and the exact position of the infrared gas leakage.

[0041] Preferably, in step S52,

[0042] According to the actual situation on site, if the infrared video with leaking gas and background color is similar, the confidence threshold is high, so the threshold is set to 0.4≤threshold≤0.6;

[0043] According to the actual situation on site, if the concentration of leaked gas is relatively low and the leak changes slowly in the infrared video, the confidence threshold is low, so the threshold is set to 0.3≤threshold<0.4.

[0044] The present invention also discloses a gas leakage detection background model, which is applicable to the above-mentioned gas leakage detection visualization identification method, and is characterized in that: the model is:

[0045] (1)

[0046] In the formula, To estimate the density, represents the sample set, , represents background pixels, represents the foreground pixel, Indicates the number of sequential frames obtained from the video forward. The radius is The volume of the inner core of the hypersphere, is a kernel function, if ,but , otherwise 0; is the diameter of the hypersphere; Represents the nuclear volume The number of samples contained in; the diameter of the hypersphere The size of is related to the sample size; with the current pixel sample as the center, the diameter of the hypersphere is gradually expanded until it contains samples; fixed nearest neighbors, take .

[0047] The present invention also relates to a pre-trained weight model for gas leakage detection, which is suitable for the above-mentioned gas leakage detection visualization identification method, and is characterized in that it includes two layers of network structures composed of convolution Conv, batch normalization BN and activation function SiLU, spliced ​​with a yolov5 network model structure; one layer is used to input a foreground pixel image, and the other layer is used to input an infrared image. The two-layer network structure fuses the convolution features of the foreground pixel image and the infrared image by Add and splices them into the yolov5 network model structure.

[0048] The technical solution provided by this application has at least the following technical effects or advantages:

[0049] 1. The present invention extracts foreground motion features through background modeling methods, combines infrared image features, and uses deep learning methods to transform and train network models. It not only retains the time series feature information of video infrared images, but also has the advantage of deep learning models to fully explore the spatial features of infrared images, thereby improving model detection efficiency in a parallel manner.

[0050] 2. Few-shot learning is a branch of machine learning that focuses on the ability to learn and predict with a small number of labeled samples. Traditional machine learning and deep learning usually require a large amount of labeled data for training, but in many practical application scenarios, especially infrared gas leak detection scenarios, it is difficult and expensive to obtain a large amount of labeled data. In this case, Few-shot learning provides an alternative.

[0051] 3. The present invention does not rely on a large amount of scene data. The fused and modified yolov5 network model structure can better retain the motion characteristics between the previous and next frames. The large amount of existing data used for pre-training contains many common scenes, so the pre-trained model obtained by training has strong generalization. When used in different project scenes, only a small amount of data of the current specific scene needs to be collected for fast Few-Shot learning to achieve high accuracy.

[0052] 4. In terms of detection effect, traditional machine learning methods and detection methods that only use deep learning methods can only achieve relatively good results when the background is fixed or the background changes very little, or the detection effect is very poor, a large number of false detections occur, and they are greatly affected by the environment. This method uses image enhancement technology to preprocess the image, uses background modeling technology to obtain the foreground motion image, and then sends the preprocessed image and the foreground motion image to the modified yolov5 network model structure for pre-training. The model obtained by pre-training achieves the effect of fusion motion features. It can have the current picture features and the motion features of the previous frame at the same time when the picture difference is very small, and complete the effective identification of the leaked gas.

[0053] 5. This method combines pre-training and Few-Shot learning to ensure that the detection effect is not affected by scene changes and environmental changes. It has good generalization performance and can maintain a high detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 The following is a flowchart of the method according to the embodiment of the present invention;

[0055] Figure 2 A network model structure diagram of the method according to an embodiment of the present invention;

[0056] Figure 3 , Figure 4 This is a rendering of the infrared gas leakage actually detected by the present invention in an industrial production task. DETAILED DESCRIPTION

[0057] The present invention first collects infrared gas leakage sample data, extracts frames and annotates the data. De-noising is performed using bilateral filtering, the contrast of the image is improved by CLAHE image enhancement technology, and background subtraction is used to model the background to obtain a moving image of the foreground target. Then, the yolov5 network model structure is modified, the convolution features of the infrared image and the foreground motion image obtained above are fused by the Add method, and a large amount of infrared gas leakage data is used for pre-training to generate a pre-training weight model. Finally, a small amount of sample data collected in the current application scenario is sent to the modified yolov5 network, and the Few-Shot fine-tuning training of the model is performed to obtain an infrared gas leakage detection model suitable for the current scenario.

[0058] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0059] Example 1

[0060] This embodiment provides a gas leak detection visualization recognition method based on motion feature fusion. Figure 1 As shown, the following steps are included:

[0061] S1: Infrared gas leakage image sample data collection;

[0062] S2: Infrared gas leakage image sample data frame extraction and positive and negative sample annotation;

[0063] S3: Image preprocessing, extracting motion features of gas leakage in the image;

[0064] S4: Based on the yolov5 infrared image detection method and a large number of pre-collected infrared gas leakage data sets, the pre-trained weight model is generated in advance by fusing the annotated data and motion features; based on the pre-trained model, the Few-Shot learning of the current scene is performed to obtain a fine-tuning model for infrared gas leakage detection;

[0065] S5: Use the infrared gas leakage video to perform inference test on the trained model and output the gas leakage prediction result.

[0066] Preferably, in step S1, the image sample data acquisition includes the following steps:

[0067] S11: Prepare infrared camera, tripod, power supply and other equipment;

[0068] S12: Determine the position and angle of the detection target and set up the infrared camera;

[0069] S13: Start the infrared camera, collect the required infrared video, and save it to the specified location.

[0070] Preferably, in step S2, the image sample data extraction and labeling includes the following steps:

[0071] S21: performing frame extraction processing on all infrared gas leakage videos collected by the infrared camera to obtain infrared gas leakage images;

[0072] S22: Use the labelImg tool to draw and annotate all infrared gas leakage images with rectangular boxes of leaking gas and easily confused background to obtain positive and negative sample annotation data.

[0073] Preferably, in step S3, image preprocessing and motion feature extraction include the following steps:

[0074] S31: Use opencv image processing technology, bilateral filtering for denoising, and CLAHE image enhancement technology to improve image contrast;

[0075] S32: Select the first T frames of the video sequence and construct a background model based on the K-Nearest Neighbors (KNN) algorithm. The construction process is as follows:

[0076] (1)

[0077] In the formula, To estimate the density. represents the sample set, . represents background pixels, represents the foreground pixel, Indicates the number of sequential frames obtained from the video forward. The radius is The volume of the inner core of the hypersphere. is a kernel function, if ,but , otherwise 0. is the diameter of the hypersphere. Represents the nuclear volume The number of samples contained in the hypersphere. The size of is related to the sample size. Taking the current pixel sample as the center, gradually expand the diameter of the hypersphere until it contains samples. Fixed The nearest neighbor is usually .

[0078] S33: At time t, the background model can determine whether each new pixel sample x belongs to the background or the foreground. The evaluation process is as follows:

[0079] (2)

[0080] in, To estimate the density, and and Directly proportional relationship. It can be approximated as:

[0081] (3)

[0082] Where B is the background model weight vector. At time m, use represents the foreground of the sample, Represents the background of the sample.

[0083] S34: Update and B. Update by adding the current time sample and eliminating the oldest time sample By judging Whether it belongs to the background, to update B.

[0084] S35: Through the foreground detection algorithm based on KNN, a foreground image after removing the background can be obtained, and then an image opening operation is used to remove noise, thereby obtaining motion feature information.

[0085] Preferably, in step S4, the feature fusion and Few-Shot learning include the following steps:

[0086] S41: In order to combine the features extracted from the foreground motion image and the infrared image, this embodiment designs a feature fusion module, such as Figure 2 As shown in the figure, the two inputs on the far left are the foreground motion image and the infrared image. CBS_2 is a two-layer network structure composed of convolution Conv, batch normalization BN and activation function SiLU. The convolution features of the two images are fused by Add, and then the yolov5 network model structure is spliced ​​to form a new network model structure that can simultaneously receive two parameters of infrared image and foreground motion image.

[0087] S42: hundreds of thousands of infrared gas leakage data sets collected in advance are processed in step S2 to obtain foreground motion images, and processed in step S3 to obtain labeled images, and are simultaneously sent to the new network model structure obtained in S41 to train and generate a general pre-trained weight model;

[0088] S43: a small amount of sample data collected in the current application scenario is processed and fused in the same way as the pre-training data, and is sent to the new network model structure obtained in S41 to perform Few-Shot fine-tuning training of the model to obtain a detection model suitable for the current scenario;

[0089] S44: When the detection model is inferred, the foreground motion image is calculated in real time according to the current video frame image, and then the rectangular frame and confidence of the gas leakage position in the current frame image are inferred. By setting a suitable threshold (the threshold range is generally preset between 0.3 and 0.6 according to the actual situation on site), it is determined whether an infrared gas leakage occurs and the exact location of the infrared gas leakage.

[0090] Preferably, in step S5, the model reasoning test comprises the following steps:

[0091] S51: Input the infrared gas leakage video, perform frame processing, perform background subtraction modeling, and extract motion feature information;

[0092] S52: Use the new network structure obtained in step S41 as the inference network structure, and the fine-tuned detection model obtained in step S43 as the inference model, read the input frame data and motion feature information, infer the gas leakage position rectangle and confidence in the current frame image, compare with the preset threshold, determine whether infrared gas leakage occurs, and the exact location of the infrared gas leakage.

[0093] According to the actual situation on site, if the infrared video of the leaking gas is similar to the background color, the confidence threshold is high and the threshold can be set to 0.4≤threshold≤0.6. The detection effect is as shown in the attached figure. Figure 3 According to the actual situation on site, if the infrared video of the leaked gas does not change significantly (the leaked gas concentration is relatively light and the leak changes slowly), the confidence threshold is low and the threshold can be set to 0.3≤threshold<0.4. The detection effect is as shown in the attached figure. Figure 4 shown. Figure 3 and Figure 4 It indicates that the leaked gas in the device has been detected successfully and a rectangular frame is drawn. Figure 3 The confidence level of the detection results is relatively high, and it can effectively distinguish backgrounds with similar colors and reduce background interference. Figure 4 It can detect slowly changing gas leaks and effectively detect small amounts of gas leaks before the gas leak becomes larger, so that customers can be informed in advance to take protective measures and avoid large losses.

[0094] The above description is only a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements without departing from the principle of the present invention. These improvements should also be regarded as within the scope of protection of the present invention.

Claims

1. A visual recognition method for gas leak detection based on motion feature fusion, characterized in that: include: Step S1: Collect infrared gas leakage image sample data; Step S2: extracting frames and labeling positive and negative samples of the infrared gas leakage image sample data; Step S3: image preprocessing, extracting motion features of gas leakage in the image; Step S4: Fusing the annotated data and motion features, based on the YOLOv5 infrared image detection method, a large number of pre-collected infrared gas leakage data sets are used to train and generate a pre-trained weight model; based on the pre-trained weight model, Few-Shot learning of the current scene is performed to obtain a fine-tuning model for infrared gas leakage detection; Step S5: Use the trained fine-tuning model for infrared gas leakage video inference test and output the gas leakage prediction result.

2. The method for visual identification of gas leakage detection based on motion feature fusion according to claim 1 is characterized in that: In step S2, the frame extraction and labeling of the infrared gas leakage image sample data includes the following steps: S21: performing frame extraction processing on all infrared gas leakage videos collected by the infrared camera to obtain infrared gas leakage images; S22: Use the labelImg tool to draw and annotate all infrared gas leakage images with rectangular boxes of leaking gas and easily confused background to obtain positive and negative sample annotation data.

3. The method for visual identification of gas leakage detection based on motion feature fusion according to claim 1 is characterized in that: Step S3 includes: S31: Use opencv image processing technology and bilateral filtering to perform denoising, and then use CLAHE image enhancement technology to improve the image contrast; S32: Select the first T frames of the video sequence and construct a background model based on the K-Nearest Neighbors algorithm. The method for constructing the background model is as follows: （1） In the formula, To estimate the density, represents the sample set, , represents background pixels, represents the foreground pixel, Indicates the number of sequential frames obtained from the video forward. The radius is The volume of the inner core of the hypersphere, is a kernel function, if ,but , otherwise 0, is the diameter of the hypersphere, Represents the nuclear volume The number of samples contained in the hypersphere, the diameter The size of is related to the sample size. The diameter of the hypersphere is gradually expanded with the current pixel sample as the center. until it contains samples, fixed nearest neighbors; S33: At time t, the background model can determine whether each new pixel sample x belongs to the background or the foreground. The evaluation process is as follows: （2） in, To estimate the density, represents the estimated density threshold, and and Directly proportional relationship; Approximately: （3） in, is the background model weight vector, at time m, using represents the foreground of the sample, Indicates the background of the sample; S34: Update and ; Update by adding the current time sample and eliminating the oldest time sample , by judging Is it background? Update ; S35: Obtain a foreground image after removing the background through the foreground detection algorithm based on K-Nearest Neighbors, and then use an image opening operation to remove noise, thereby obtaining motion feature information.

4. The method for visual identification of gas leakage detection based on motion feature fusion according to claim 1 is characterized in that: The step S4 comprises: S41: The convolution features extracted from the foreground motion image and the infrared image are fused into a two-layer network structure consisting of convolution Conv, batch normalization BN and activation function SiLU by adding, and then the yolov5 network model structure is spliced ​​to form a new network model structure that can simultaneously receive two parameters of the infrared image and the foreground motion image; S42: several infrared gas leakage data sets collected in advance are processed in step S2 to obtain foreground motion images, and processed in step S3 to obtain labeled images, and are simultaneously sent to the new network model structure described in step S41 to train and generate a universal pre-trained weight model; S43: A small amount of sample data collected for the current application scenario is subjected to the same data processing and data fusion method as the pre-trained data in step S42, and then sent to the new network model structure described in step S41 to perform Few-Shot fine-tuning training of the model to obtain a fine-tuned detection model suitable for the current scenario.

5. The method for visual identification of gas leakage detection based on motion feature fusion according to claim 4 is characterized in that: In step S5, the model is subjected to inference testing, including the following steps: S51: Input the infrared gas leakage video, perform frame processing, perform background subtraction modeling, and extract motion feature information; S52: Use the new network model structure obtained in S41 as the inference network structure, and the fine-tuned detection model obtained in S43 as the inference model, read the input frame data and motion feature information, infer the gas leakage position rectangle and confidence in the current frame image, compare with the preset threshold, and determine whether infrared gas leakage occurs and the exact position of the infrared gas leakage.

6. The method for visual identification of gas leakage detection based on motion feature fusion according to claim 5 is characterized in that: In step S52, According to the actual situation on site, if the infrared video with leaking gas and background color is similar, the confidence threshold is high, so the threshold is set to 0.4≤threshold≤0.6; According to the actual situation on site, if the concentration of leaked gas is relatively low and the leak changes slowly in the infrared video, the confidence threshold is low, so the threshold is set to 0.3≤threshold<0.

4.

7. A gas leak detection background model, applicable to the gas leak detection visualization identification method according to any one of claims 1 to 6, characterized in that: The model is: （1） In the formula, To estimate the density, represents the sample set, , represents background pixels, represents the foreground pixel, Indicates the number of sequential frames obtained from the video forward. The radius is The volume of the inner core of the hypersphere, is a kernel function, if ,but , otherwise 0; is the diameter of the hypersphere; Represents the nuclear volume The number of samples contained in; the diameter of the hypersphere The size of is related to the sample size; with the current pixel sample as the center, the diameter of the hypersphere is gradually expanded until it contains samples; fixed nearest neighbors, take .

8. A pre-trained weight model for gas leak detection, applicable to the gas leak detection visualization identification method according to any one of claims 1 to 6, characterized in that: The method comprises two layers of network structures consisting of convolution Conv, batch normalization BN and activation function SiLU, and splicing the yolov5 network model structure; one layer is used to input a foreground pixel image, and the other layer is used to input an infrared image. The two-layer network structure fuses the convolution features of the foreground pixel image and the infrared image by Add and splices them into the yolov5 network model structure.

Images