Temperature-aware laboratory alerting method and laboratory alerting device

Abstract

The embodiment of the application discloses a laboratory early warning method and device based on temperature sensing, relates to the technical field of laboratory monitoring, and comprises the following steps: acquiring a laboratory image segmentation model based on day and night images; acquiring laboratory monitoring images in real time, wherein the laboratory monitoring images comprise thermal imaging images; judging whether there is a high-temperature danger in the current laboratory based on the thermal imaging images; if yes, segmenting the laboratory monitoring images based on the laboratory image segmentation model to generate laboratory segmentation images; and generating corresponding early warning information based on the thermal imaging images and the laboratory segmentation images. Through the all-day heat source influence monitoring of the laboratory based on the deep learning model, the use safety of the laboratory is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of laboratory monitoring technology, and more specifically to a laboratory early warning method and a laboratory early warning device based on temperature sensing. Background Technology

[0002] In recent years, alarms for combustion and explosion situations in laboratories have received increasing attention.

[0003] Traditional alarm methods rely on smoke detectors to sense smoke in the event of a fire in a laboratory and trigger an alarm accordingly. However, this method has two main drawbacks: it requires a large amount of smoke to be detected, and it cannot detect explosions or other similar situations in advance, thus failing to meet practical needs.

[0004] With the continuous development of deep learning technology, people are considering using deep learning to segment and identify image datasets containing elements such as smoke and flames in laboratories, so as to immediately trigger an alarm when smoke or flames occur in the laboratory. However, on the one hand, traditional deep learning relies on a large number of labeled images, which results in a large workload and a great burden on technicians; on the other hand, the above methods still have a lag, that is, they can only identify and alarm after the emergency has occurred, which often makes it impossible to take corresponding fire-fighting measures in time, which can easily lead to greater losses. Summary of the Invention

[0005] In order to overcome the above-mentioned technical problems in the prior art, the present invention provides a laboratory early warning method and a laboratory early warning device based on temperature sensing. By using a deep learning model to monitor the heat source impact on the laboratory throughout the day, the safety of laboratory use is effectively improved.

[0006] To achieve the above objectives, embodiments of the present invention provide a laboratory early warning method based on temperature sensing. The method includes: acquiring a laboratory image segmentation model based on day and night images; acquiring laboratory monitoring images in real time, wherein the laboratory monitoring images are thermal imaging images; determining whether there is a high-temperature hazard in the laboratory based on the laboratory monitoring images; if so, segmenting the laboratory monitoring images based on the laboratory image segmentation model to generate a laboratory segmented image; and generating corresponding early warning information based on the laboratory monitoring images and the laboratory segmented images.

[0007] Preferably, the method further includes: before obtaining the laboratory image segmentation model based on day and night images, obtaining a teacher model and a student model; obtaining a preset day and night image pair dataset, calibrating the preset day and night image pair dataset to obtain a calibrated dataset; analyzing the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; creating a day and night discrimination network, optimizing the student model based on the day and night discrimination network to obtain an optimized student model; training the optimized student model based on the label information to generate a trained student model; and updating the teacher model based on the trained student model to generate the laboratory image segmentation model.

[0008] Preferably, the step of generating corresponding early warning information based on the laboratory monitoring image and the laboratory segmentation image includes: determining heat source information based on the laboratory monitoring image, the heat source information including the heat source heating type and heat source temperature; determining the heat source item type and adjacent items within a preset range of the heat source based on the laboratory segmentation image; judging whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the heat source item type, and the item information of the adjacent items, and generating a judgment result; and generating corresponding early warning information based on the judgment result.

[0009] Preferably, the step of determining whether the heat source and the adjacent items pose a safety risk based on the heat source information, the type of the heat source item, and the item information of the adjacent items, and generating a determination result, includes: determining a first combustion / explosion threshold value corresponding to the heat source item based on the heat source item type; determining whether the heat source item poses a safety risk based on the heat source temperature and the first combustion / explosion threshold value, and generating a first determination result; determining a second combustion / explosion threshold value of the adjacent items based on the item information; determining whether the adjacent items pose a safety risk based on the heat source temperature, the heat source heating type, and the second combustion / explosion threshold value, and generating a second determination result; and generating a determination result indicating whether the laboratory poses a safety risk based on the first determination result and the second determination result.

[0010] Preferably, the item information includes the temperature rise information of adjacent items, and the step of generating corresponding early warning information based on the judgment result includes: if the judgment result indicates that the heat source item has a safety risk and / or the adjacent item has a safety risk, generating corresponding first early warning information; if the judgment result indicates that neither the heat source item nor the adjacent item has a safety risk: determining a first temperature change curve of the heat source item based on the real-time acquired heat source temperature; generating a second temperature change curve of the adjacent item based on the first temperature change curve, the heat source heating type, and the temperature rise information; generating first prediction information based on the first temperature change curve and the first combustion / explosion threshold; generating second prediction information based on the second temperature change curve and the second combustion / explosion threshold; generating corresponding second early warning information based on the first prediction information; and generating corresponding third early warning information based on the second prediction information.

[0011] Accordingly, the present invention also provides a laboratory early warning device based on temperature sensing. The device includes: a model acquisition unit for acquiring a laboratory image segmentation model based on day and night images; an image acquisition unit for acquiring laboratory monitoring images in real time, wherein the laboratory monitoring images are thermal imaging images; a judgment unit for judging whether there is a high temperature hazard in the current laboratory based on the laboratory monitoring images; an image segmentation unit for segmenting the laboratory monitoring images based on the laboratory image segmentation model to generate a laboratory segmented image if the hazard is found; and an early warning unit for generating corresponding early warning information based on the laboratory monitoring images and the laboratory segmented images.

[0012] Preferably, the apparatus further includes a model creation unit, which is configured to: acquire a teacher model and a student model before acquiring the laboratory image segmentation model based on day and night images; acquire a preset day and night image pair dataset, calibrate the preset day and night image pair dataset to obtain a calibrated dataset; analyze the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; create a day and night discrimination network, optimize the student model based on the day and night discrimination network to obtain an optimized student model; train the optimized student model based on the label information to generate a trained student model; and update the teacher model based on the trained student model to generate the laboratory image segmentation model.

[0013] Preferably, the early warning unit includes: a heat source information determination module, used to determine heat source information based on the laboratory monitoring image, the heat source information including the heat source heating type and heat source temperature; an adjacent information determination module, used to determine the heat source item type and adjacent items within a preset range of the heat source based on the laboratory segmentation image; a risk judgment module, used to judge whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the heat source item type, and the item information of the adjacent items, and generate a judgment result; and an early warning module, used to generate corresponding early warning information based on the judgment result.

[0014] Preferably, the risk assessment module is specifically used for: determining a first combustion / explosion threshold value corresponding to the heat source item based on the heat source item type; determining whether the heat source item poses a safety risk based on the heat source temperature and the first combustion / explosion threshold value, and generating a first assessment result; determining a second combustion / explosion threshold value of the adjacent item based on the item information; determining whether the adjacent item poses a safety risk based on the heat source temperature, the heat source heating type, and the second combustion / explosion threshold value, and generating a second assessment result; and generating a assessment result regarding whether the laboratory poses a safety risk based on the first assessment result and the second assessment result.

[0015] Preferably, the item information includes the temperature rise information of adjacent items, and the early warning module is specifically used for: if the judgment result indicates that the heat source item has a safety risk and / or the adjacent items have a safety risk, generating corresponding first early warning information; if the judgment result indicates that neither the heat source item nor the adjacent items have a safety risk: determining a first temperature change curve of the heat source item based on the real-time acquired heat source temperature; generating a second temperature change curve of the adjacent items based on the first temperature change curve, the heat source heating type, and the temperature rise information; generating first prediction information based on the first temperature change curve and the first combustion / explosion threshold; generating second prediction information based on the second temperature change curve and the second combustion / explosion threshold; generating corresponding second early warning information based on the first prediction information; and generating corresponding third early warning information based on the second prediction information.

[0016] The present invention has at least the following technical effects through the technical solution provided by the present invention:

[0017] By combining the actual needs of laboratory monitoring, a laboratory image segmentation model based on day and night images is trained, which enables accurate and effective segmentation of laboratory monitoring images throughout the day, ensuring the accuracy of subsequent heat source monitoring and early warning information.

[0018] On the other hand, by monitoring the temperature / heating of heat sources and adjacent items based on the actual situation of laboratory safety risks, a comprehensive analysis can be conducted to determine whether there are any safety risks in the laboratory. This effectively ensures the accuracy and effectiveness of laboratory safety monitoring and improves the reliability of early warning.

[0019] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0021] Figure 1 This is a flowchart illustrating the specific implementation of the temperature-sensing-based laboratory early warning method provided in this embodiment of the invention.

[0022] Figure 2 This is a flowchart illustrating the specific implementation of generating early warning information based on laboratory monitoring images and laboratory segmentation images provided in this embodiment of the invention.

[0023] Figure 3 This is a flowchart illustrating the specific implementation of determining whether there is a safety risk between a heat source and adjacent items, as provided in an embodiment of the present invention.

[0024] Figure 4 This is a schematic diagram of the structure of a temperature-sensing-based laboratory early warning device provided in an embodiment of the present invention. Detailed Implementation

[0025] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0026] In this invention, the terms "system" and "network" are used interchangeably. "Multiple" refers to two or more; therefore, in this invention, "multiple" can also be understood as "at least two." "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / ", unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, it should be understood that in the description of this invention, terms such as "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order.

[0027] The background technology of this invention will be introduced first below.

[0028] Currently, there is no effective automatic sensing and identification scheme for early warning of combustion or explosion in laboratories. Traditional deep learning-based image recognition methods can identify smoke, sparks, and fires. However, these methods are often designed for specific scenarios, primarily daytime environments (where lighting changes are minimal). Laboratories, on the other hand, require 24-hour monitoring, which traditional methods cannot meet. Furthermore, traditional image recognition methods require extensive labeling, resulting in a huge workload and thus failing to meet practical needs.

[0029] To resolve the above technical issues, please refer to [link / reference]. Figure 1 This invention provides a laboratory early warning method based on temperature sensing, the method comprising:

[0030] S10) Obtain a laboratory image segmentation model based on day and night images;

[0031] S20) Acquire laboratory monitoring images in real time, wherein the laboratory monitoring images are thermal imaging images;

[0032] S30) Determine whether there is a high-temperature hazard in the laboratory based on the laboratory monitoring images;

[0033] S40) If so, segment the laboratory monitoring image based on the laboratory image segmentation model to generate a laboratory segmented image;

[0034] S50) Generate corresponding early warning information based on the laboratory monitoring image and the laboratory segmentation image.

[0035] In one possible implementation, a laboratory image segmentation model based on day and night images is first obtained. For example, this model is pre-trained and generated based on a day and night image dataset. Specifically, in this embodiment of the invention, the method further includes: before obtaining the laboratory image segmentation model based on day and night images, obtaining a teacher model and a student model; obtaining a preset day and night image pair dataset, calibrating the preset day and night image pair dataset to obtain a calibrated dataset; analyzing the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; creating a day and night discrimination network, optimizing the student model based on the day and night discrimination network to obtain an optimized student model; training the optimized student model based on the label information to generate a trained student model; and updating the teacher model based on the trained student model to generate the laboratory image segmentation model.

[0036] For example, in one embodiment, a teacher-student model is used for training. First, an initial teacher model and student model are obtained. For instance, a PSPNet for segmenting RGB images can be pre-trained on the Mapillary Vistas dataset (which contains only RGB images and their semantic annotations), and this model can be used as the teacher model. This provides segmentation labels for all daytime images subsequently trained on another dataset.

[0037] Then, a preset day-night RGB-T image pair is obtained as the dataset (e.g., the Freiburg Thermal dataset, which contains several RGB-T image pairs). Since the images are formed by taking pictures during the day and at night respectively, this day-night image pair dataset needs to be calibrated to improve the accuracy of subsequent training. Specifically, the day-night image pairs in the dataset can be calibrated based on camera calibration methods. For example, the intrinsic parameter matrices can be obtained from the internal calibration of the RGB and thermal imaging cameras respectively. and distortion coefficient Based on these parameters, the lossless intrinsic mechanical parameter matrices of the two cameras are obtained. and Then, the RGB image depth map obtained from depth estimation is used. All RGB pixel coordinates are mapped to 3D space, and then these 3D coordinates are projected onto the thermal image. Specifically, for a 2D coordinate i, let... Points in an undistorted RGB image If the depth value is at point i, then it can be obtained by inverting the camera intrinsic parameter matrix. Mapped to 3D space, this is specifically represented as:

[0038]

[0039] Then use the calibrated extrinsic matrix of the thermal imaging camera. and Projecting 3D coordinates onto a thermal imaging image can be specifically characterized as follows:

[0040]

[0041] in Its mapping method is the same as that of RGB images. Thus, RGB-T alignment can be completed by projecting the RGB image onto the thermal image, allowing the semantic annotations on the RGB image to be directly applied to the corresponding thermal image. At this point, based on the teacher model... The daytime images in the above dataset are analyzed to generate corresponding label information, which can then be used to train the student model.

[0042] Because of the teacher model's supervision, the student model did not generalize to nighttime image segmentation. Therefore, to improve the model's accuracy in recognizing and segmenting day and night images, a domain-adaptive approach is used to optimize the student model. Specifically, a day-night discrimination network C can be created and then inserted into the student model. After the softmax prediction layer, to generate the optimized student model, the discrimination network... Segmentation results or ( If the input image is daytime, then the output image is denoted as... Otherwise, record as As input, it is used to identify whether the image to be analyzed is from daytime or nighttime. For example, the loss function of this day / night discrimination network can be characterized as:

[0043] .

[0044] At this point, the multimodal RGB-T model is tested using the aforementioned calibrated dataset. (The "student" model) undergoes supervised training by minimizing the cross-entropy loss. Specifically, the optimized student model is trained in a supervised manner based on the aforementioned label information, and its loss function can be represented as:

[0045]

[0046] in, Teacher model for RGB images The prediction results The student model is represented as a pair of RGB-T images ( The prediction results.

[0047] Although thermal images remain largely unchanged regardless of daytime lighting changes, RGB images exhibit significant differences between day and night, displaying substantial domain gaps. Therefore, the network parameters for daytime and nighttime images are not shared but processed separately. An alternating training strategy is adopted: the parameters of the segmentation model are frozen during the training of the discriminator model; the parameters of the discriminator model are frozen during the training of the whole segmentation model. These two processes are gradually alternated. During iterative training, each iteration inputs a pair of daytime RGB-T images and a pair of nighttime RGB-T images. Specifically, a two-step strategy is employed: the first step involves separately training the semantic segmentation network for the daytime domain. Its loss function can be characterized as: 7) The second step is to train the discriminator network C separately, whose loss function can be represented as: After performing multiple rounds of iterative training and achieving the expected results, the teacher model is updated based on the trained student model to generate a laboratory image segmentation model based on day and night images.

[0048] In this embodiment of the invention, by training a laboratory image segmentation model based on day and night images, the internal feature distribution of the multimodal segmentation network can be calibrated using domain adaptation technology, enabling the model to have the same recognition and segmentation accuracy for both night and day images, thereby greatly improving the model's adaptability and meeting practical needs.

[0049] After creating the aforementioned laboratory image segmentation model, during laboratory monitoring, real-time laboratory monitoring images are acquired, such as thermal imaging images obtained through a thermal imaging camera. Based on these images, it is determined whether the laboratory poses a high-temperature hazard. Specifically, different temperatures in the monitoring images can be labeled with different colors. During the judgment, an RGB model is used to segment the colors within the original thermal image, obtaining different color regions. For example, based on different temperatures, nine color regions could be defined, each corresponding to a temperature range. When a color region with a temperature exceeding a certain threshold is found, it is determined that the region poses a high-temperature hazard.

[0050] At this point, the laboratory monitoring image is segmented based on the aforementioned laboratory image segmentation model, generating a corresponding segmented laboratory image. Then, the laboratory monitoring image and the segmented laboratory image are used to generate corresponding early warning information. For details, please refer to [link to relevant documentation]. Figure 2 In this embodiment of the invention, generating corresponding early warning information based on the laboratory monitoring image and the laboratory segmentation image includes:

[0051] S51) Determine heat source information based on the laboratory monitoring images, wherein the heat source information includes the heat source heating type and the heat source temperature;

[0052] S52) Determine the type of heat source item and adjacent items within a preset range of the heat source based on the laboratory segmentation image;

[0053] S53) Based on the heat source information, the type of heat source item, and the item information of the adjacent items, determine whether there is a safety risk to the heat source and the adjacent items, and generate a judgment result;

[0054] S54) Generate corresponding early warning information based on the judgment result.

[0055] In one possible implementation, the heat source information is first determined based on laboratory monitoring images. Specifically, the location of the heat source is determined, along with information including but not limited to the heating type and temperature of the heat source. The heating type includes direct heating (combustion heating, chemical exothermic reaction) and indirect heating (radiative heating). Then, the type of heat source item and adjacent items within a preset range near the heat source are determined by segmenting the laboratory images. For example, in one embodiment, an experimenter accidentally drips a chemical reagent onto a paper towel on the test bench, causing a chemical reaction that releases a large amount of heat. At this time, the heat source is detected as the paper towel (flammable item). The only adjacent item near the paper towel is the test bench, and the heat source is basically in contact with the test bench. Therefore, it can be determined that the heating type of the paper towel and the test bench is direct heating, and the current temperature of the heat source has reached 120°C.

[0056] At this point, based on the aforementioned heat source information, heat source item type, and adjacent item information, to determine whether there is a safety risk to the heat source and adjacent items, please refer to [link to relevant documentation]. Figure 3 In this embodiment of the invention, the step of determining whether there is a safety risk between the heat source and the adjacent items based on the heat source information, the type of the heat source item, and the item information of the adjacent items, and generating a determination result, includes:

[0057] S531) Determine a first combustion / explosion threshold value corresponding to the heat source item based on the type of heat source item;

[0058] S532) Based on the heat source temperature and the first combustion / explosion threshold, determine whether the heat source item poses a safety risk, and generate a first judgment result;

[0059] S533) Determine the second combustion / explosion threshold value of the adjacent items based on the item information;

[0060] S534) Based on the heat source temperature, the heat source heating type and the second combustion / explosion threshold, determine whether there is a safety risk to the adjacent items, and generate a second judgment result;

[0061] S535) Based on the first judgment result and the second judgment result, a judgment result is generated to determine whether there is a safety risk in the laboratory.

[0062] For example, in one embodiment, the heat source is a chemical solution contained in a container. This chemical solution needs to be stored in the absence of air, otherwise it will heat up rapidly due to the action of air. At the same time, there is another sealed container containing flammable gas near the heat source. Due to improper operation by the experimenter, the container containing the chemical solution was forgotten to be closed after use, causing the temperature of the container to rise rapidly. On the one hand, the container and the test bench are rapidly heated, and on the other hand, the container containing the flammable gas is radiated and heated. During the monitoring process, the first combustion critical value (which is easily known to those skilled in the art, the parameter for this container can be the first (due to high temperature) burn-out critical value) is determined to be 500°C based on the type of container containing the solution. The first combustion critical value of the test bench is 1000°C. At this time, the temperature of the heat source is monitored to have reached 400°C (for example, the threshold of triggering 80% of the first combustion critical value). Therefore, it can be determined that the container containing the solution has a safety risk, while the test bench does not have a safety risk.

[0063] On the other hand, the second combustion / explosion criticality value of the container storing the flammable gas is further obtained. For example, if the container's second combustion criticality value is 500°C and its explosion criticality value is 300°C, and the container has already been radiated to a temperature of 50°C, it can be determined that the container currently poses no safety risk. Combining the above first and second judgment results, a judgment result on whether there is a safety risk in the current laboratory can be generated. For example, if either of the above two judgment results indicates the presence of a safety risk, then it is determined that there is a safety risk in the current laboratory. At this time, corresponding early warning information can be generated based on the above judgment result, and the relevant management personnel can be notified immediately.

[0064] In this embodiment of the invention, during the process of safety monitoring of the laboratory, since the risk of combustion / explosion is often closely related to the items around the heat source, not only is the temperature of the heat source in the laboratory monitored, but also the temperature of related items near the heat source is monitored. This allows for a comprehensive assessment of whether there is a safety risk in the laboratory under the influence of the heat source, thereby effectively improving the effectiveness and accuracy of safety monitoring of the laboratory.

[0065] However, in practical applications, if the corresponding warning is only issued when the heat source has already triggered the critical condition, on the one hand, the managers may not have enough time to take corresponding actions, which may still lead to combustion / explosion accidents; on the other hand, if the monitoring and early warning are not carried out at the beginning of the heat source temperature rise, the opportunity for a more reasonable alarm is missed, so the warning effect cannot meet the higher requirements of the managers.

[0066] To address the aforementioned technical problems, in this embodiment of the invention, the item information includes the heating temperature rise information of adjacent items. The step of generating corresponding early warning information based on the judgment result includes: if the judgment result indicates that the heat source item has a safety risk and / or the adjacent items have a safety risk, generating corresponding first early warning information; if the judgment result indicates that neither the heat source item nor the adjacent items have a safety risk: determining a first temperature change curve of the heat source item based on the real-time acquired heat source temperature; generating a second temperature change curve of the adjacent items based on the first temperature change curve, the heat source heating type, and the heating temperature rise information; generating first prediction information based on the first temperature change curve and the first combustion / explosion threshold; generating second prediction information based on the second temperature change curve and the second combustion / explosion threshold; generating corresponding second early warning information based on the first prediction information; and generating corresponding third early warning information based on the second prediction information.

[0067] In one possible implementation, after obtaining the above-mentioned assessment result regarding whether there is a safety risk in the current laboratory, if at least one of the heat source item or its adjacent items poses a safety risk, a corresponding first warning message is immediately generated to remind management personnel to take corresponding measures or conduct real-time monitoring to avoid an accident. If the assessment result indicates that neither the heat source item nor its adjacent items pose a safety risk, then the safety risks of the heat source item and the adjacent items are tracked and monitored separately.

[0068] Specifically, firstly, a first temperature change curve for the heat source item is determined based on the real-time acquired heat source temperature. Then, based on the first temperature change curve, combined with information such as the heating type of the heat source and the heating rate of adjacent items, a second temperature change curve for the adjacent items is generated. Based on the first temperature change curve and the first combustion / explosion critical value of the heat source, the time that the heat source item may need to reach the safety risk threshold is calculated, and corresponding first prediction information is generated. Similarly, based on the second temperature change curve and the second combustion / explosion critical value of the adjacent items, the time that the adjacent items may need to reach the safety risk threshold is calculated, and corresponding second prediction information is generated. Finally, corresponding early warning information is generated based on the aforementioned prediction information.

[0069] During the monitoring process, managers can quantitatively judge whether there is a safety risk in the laboratory and the urgency of the safety risk by the early warning information generated by the above-mentioned predictive information. They can then determine whether immediate rescue measures are needed, whether to take gradual countermeasures, or whether to simply observe. This greatly improves the timeliness of the early warning, gives managers more time to deal with any possible safety risks, significantly improves the accuracy and effectiveness of laboratory safety management, reduces the probability of safety accidents, and meets actual safety needs.

[0070] The temperature-sensing-based laboratory early warning device provided by the embodiments of the present invention will be described below with reference to the accompanying drawings.

[0071] Please see Figure 4 Based on the same inventive concept, this invention provides a temperature-sensing-based laboratory early warning device. The device includes: a model acquisition unit for acquiring a laboratory image segmentation model based on day and night images; an image acquisition unit for acquiring laboratory monitoring images in real time, wherein the laboratory monitoring images are thermal imaging images; a judgment unit for judging whether there is a high-temperature hazard in the current laboratory based on the laboratory monitoring images; an image segmentation unit for segmenting the laboratory monitoring images based on the laboratory image segmentation model to generate a laboratory segmented image if the hazard is found; and an early warning unit for generating corresponding early warning information based on the laboratory monitoring images and the laboratory segmented images.

[0072] In this embodiment of the invention, the device further includes a model creation unit, which is configured to: acquire a teacher model and a student model before acquiring the laboratory image segmentation model based on day and night images; acquire a preset day and night image pair dataset, calibrate the preset day and night image pair dataset to obtain a calibrated dataset; analyze the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; create a day and night discrimination network, optimize the student model based on the day and night discrimination network to obtain an optimized student model; train the optimized student model based on the label information to generate a trained student model; and update the teacher model based on the trained student model to generate the laboratory image segmentation model.

[0073] In this embodiment of the invention, the early warning unit includes: a heat source information determination module, used to determine heat source information based on the laboratory monitoring image, the heat source information including the heat source heating type and heat source temperature; an adjacent information determination module, used to determine the heat source item type and adjacent items within a preset range of the heat source based on the laboratory segmentation image; a risk judgment module, used to judge whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the heat source item type, and the item information of the adjacent items, and generate a judgment result; and an early warning module, used to generate corresponding early warning information based on the judgment result.

[0074] In this embodiment of the invention, the risk assessment module is specifically used for: determining a first combustion / explosion threshold value corresponding to the heat source item based on the heat source item type; determining whether the heat source item poses a safety risk based on the heat source temperature and the first combustion / explosion threshold value, and generating a first assessment result; determining a second combustion / explosion threshold value of the adjacent item based on the item information; determining whether the adjacent item poses a safety risk based on the heat source temperature, the heat source heating type, and the second combustion / explosion threshold value, and generating a second assessment result; and generating a assessment result indicating whether the laboratory poses a safety risk based on the first assessment result and the second assessment result.

[0075] In this embodiment of the invention, the item information includes the temperature rise information of adjacent items. The early warning module is specifically used for: if the judgment result indicates that the heat source item has a safety risk and / or the adjacent item has a safety risk, generating corresponding first early warning information; if the judgment result indicates that neither the heat source item nor the adjacent item has a safety risk: determining a first temperature change curve of the heat source item based on the real-time acquired heat source temperature; generating a second temperature change curve of the adjacent item based on the first temperature change curve, the heat source heating type, and the temperature rise information; generating first prediction information based on the first temperature change curve and the first combustion / explosion threshold; generating second prediction information based on the second temperature change curve and the second combustion / explosion threshold; generating corresponding second early warning information based on the first prediction information; and generating corresponding third early warning information based on the second prediction information.

[0076] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0077] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not describe the various possible combinations separately.

[0078] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0079] Furthermore, various different implementations of the present invention can be combined arbitrarily, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed in the present invention.

Claims

1. A laboratory early warning method based on temperature sensing, characterized in that, The method includes: A laboratory image segmentation model based on day and night images is obtained. The steps for obtaining the laboratory image segmentation model include: before obtaining the laboratory image segmentation model based on day and night images, obtaining a teacher model and a student model; obtaining a preset day and night image pair dataset, calibrating the preset day and night image pair dataset to obtain a calibrated dataset; analyzing the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; creating a day and night discrimination network, optimizing the student model based on the day and night discrimination network to obtain an optimized student model; training the optimized student model based on the label information to generate a trained student model; and updating the teacher model based on the trained student model to generate the laboratory image segmentation model. Real-time acquisition of laboratory monitoring images, including thermal imaging images; Based on the thermal imaging images, determine whether there is a high-temperature hazard in the current laboratory; If so, the laboratory monitoring image is segmented based on the laboratory image segmentation model to generate a segmented laboratory image; Based on the thermal imaging image and the laboratory segmentation image, corresponding early warning information is generated: Heat source information is determined based on the thermal imaging image, including the heating type and temperature of the heat source; the type of heat source item and adjacent items within a preset range of the heat source are determined based on the laboratory segmentation image; based on the heat source information, the type of heat source item, and the item information of the adjacent items, a judgment is made regarding whether there is a safety risk to the heat source and the adjacent items, and a judgment result is generated; based on the judgment result, corresponding early warning information is generated.

2. The method according to claim 1, characterized in that, The step of determining whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the type of the heat source item, and the item information of the adjacent items, and generating a determination result, includes: Determine the first combustion / explosion threshold value corresponding to the heat source item based on the type of heat source item; Based on the heat source temperature and the first combustion / explosion threshold, determine whether the heat source item poses a safety risk and generate a first judgment result; Based on the item information, determine the second combustion / explosion threshold value of the adjacent items; Based on the heat source temperature, the heat source heating type, and the second combustion / explosion threshold, a second judgment result is generated to determine whether there is a safety risk to the adjacent items. Based on the first and second judgment results, a judgment result is generated to determine whether there is a safety risk in the laboratory.

3. The method according to claim 2, characterized in that, The item information includes the temperature rise information of adjacent items, and the generation of corresponding early warning information based on the judgment result includes: If the judgment result indicates that the heat source item has a safety risk and / or the adjacent item has a safety risk, a corresponding first warning message is generated; If the judgment result indicates that neither the heat source item nor the adjacent item poses a safety risk: A first temperature change curve of the heat source item is determined based on the real-time acquired heat source temperature; A second temperature change curve for the adjacent items is generated based on the first temperature change curve, the heating type of the heat source, and the heating temperature rise information. First prediction information is generated based on the first temperature change curve and the first combustion / explosion critical value; Second prediction information is generated based on the second temperature change curve and the second combustion / explosion critical value; A second early warning message is generated based on the first prediction information, and a third early warning message is generated based on the second prediction information.

4. A laboratory early warning device based on temperature sensing, characterized in that, The device includes: The model acquisition unit is used to acquire a laboratory image segmentation model based on day and night images; An image acquisition unit is used to acquire laboratory monitoring images in real time, including thermal imaging images. The judgment unit is used to determine whether there is a high-temperature hazard in the current laboratory based on the thermal imaging image; An image segmentation unit is used, if so, to segment the laboratory monitoring image based on the laboratory image segmentation model to generate a segmented laboratory image; An early warning unit is used to generate corresponding early warning information based on the thermal imaging image and the laboratory segmentation image. The early warning unit includes a heat source information determination module, an adjacent information determination module, a risk judgment module, and an early warning module. The heat source information determination module is used to determine heat source information based on the thermal imaging image, including the heat source heating type and temperature. The adjacent information determination module is used to determine the type of heat source item and adjacent items within a preset range of the heat source based on the laboratory segmentation image. The risk judgment module is used to determine whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the type of heat source item, and the item information of the adjacent items, and generate a judgment result. The early warning module is used to generate corresponding early warning information based on the judgment result: determining heat source information based on the thermal imaging image, including the heat source heating type and temperature; determining the type of heat source item and adjacent items within a preset range of the heat source based on the laboratory segmentation image; determining whether there is a safety risk to the heat source and the adjacent items based on the heat source information, the type of heat source item, and the item information of the adjacent items, and generating a judgment result; and generating corresponding early warning information based on the judgment result. The device further includes a model creation unit, which is configured to: acquire a teacher model and a student model before acquiring the laboratory image segmentation model based on day and night images; acquire a preset day and night image pair dataset, calibrate the preset day and night image pair dataset to obtain a calibrated dataset; analyze the daytime image set in the calibrated dataset based on the teacher model to generate corresponding label information; create a day and night discrimination network, optimize the student model based on the day and night discrimination network to obtain an optimized student model; train the optimized student model based on the label information to generate a trained student model; and update the teacher model based on the trained student model to generate the laboratory image segmentation model.

5. The apparatus according to claim 4, characterized in that, The risk assessment module is specifically used for: Determine the first combustion / explosion threshold value corresponding to the heat source item based on the type of heat source item; Based on the heat source temperature and the first combustion / explosion threshold, determine whether the heat source item poses a safety risk and generate a first judgment result; Based on the item information, determine the second combustion / explosion threshold value of the adjacent items; Based on the heat source temperature, the heat source heating type, and the second combustion / explosion threshold, a second judgment result is generated to determine whether there is a safety risk to the adjacent items. Based on the first and second judgment results, a judgment result is generated to determine whether there is a safety risk in the laboratory.

6. The apparatus according to claim 5, characterized in that, The item information includes the temperature rise information of adjacent items, and the early warning module is specifically used for: If the judgment result indicates that the heat source item has a safety risk and / or the adjacent item has a safety risk, a corresponding first warning message is generated; If the judgment result indicates that neither the heat source item nor the adjacent item poses a safety risk: A first temperature change curve of the heat source item is determined based on the real-time acquired heat source temperature; A second temperature change curve for the adjacent items is generated based on the first temperature change curve, the heating type of the heat source, and the heating temperature rise information. First prediction information is generated based on the first temperature change curve and the first combustion / explosion critical value; Second prediction information is generated based on the second temperature change curve and the second combustion / explosion critical value; A second early warning message is generated based on the first prediction information, and a third early warning message is generated based on the second prediction information.

Images