Video monitoring method and device for park hazardous chemical substance warehouse, equipment and storage medium

Abstract

The invention provides a video monitoring method and device for a park hazardous chemical substance warehouse, equipment and a storage medium, and the method comprises the steps: associating at least one target hazardous chemical type for each preset warehouse identifier, determining the hazardous chemical type, needing to be monitored, of each warehouse, determining an associated edge node corresponding to each preset warehouse identifier according to the target hazardous chemical type, creating candidate containers for different warehouse identifications at the associated edge node so as to configure the same identification model, realizing video identification of a type of hazardous chemical types through one edge node, and realizing data isolation of different warehouses by utilizing the containers; after the monitoring camera is configured with the target warehouse identifier, the monitoring video stream of the same monitoring camera is automatically sent to a plurality of edge nodes in parallel according to the node mapping table, and each edge node executes one type of recognition, so that rapid recognition can be realized through a lighter recognition model, and the recognition efficiency is improved. And the response efficiency and accuracy of hazardous chemical substance monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a video monitoring method, device, equipment, and storage medium for a hazardous chemicals warehouse in a park. Background Art

[0002] At present, some large production parks are usually equipped with multiple hazardous chemical warehouses to store different production materials. Even the same hazardous chemical may correspond to multiple risk types, including common flammable and explosive, corrosive or toxic, etc. Hazardous chemical warehouses need to follow strict safety measures. Usually, multiple surveillance cameras are installed inside and outside each hazardous chemical warehouse. The relevant artificial intelligence model is used to identify the video stream, and timely warnings are issued when equipment abnormalities or risky actions of operators are detected, which can greatly improve the safety of hazardous chemical warehouses.

[0003] However, the AI ​​model needs to be trained with training data. Different risk types have different corresponding safety requirements. In corrosive areas, safety clothing needs to be worn correctly. In high-pressure gas warehouses, workers' handling postures need to be checked to see if they are compliant. In related technologies, all camera video streams are usually input into the same AI model for risk identification. Since each risk type includes multiple risk operations, the AI ​​model can identify more types of actions. The model training is more complex and the amount of calculation is very large. The long calculation time will cause the risk warning to have a certain lag, which cannot meet the monitoring needs of hazardous chemical warehouses. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a video monitoring method, device, equipment, and storage medium for a hazardous chemical warehouse in a park, which can deploy multiple recognition models through multiple edge nodes, associate cameras and edge nodes through warehouse identification, and improve the accuracy and efficiency of risk identification.

[0005] In a first aspect, an embodiment of the present invention provides a video monitoring method for a hazardous chemical warehouse in a park, which is applied to a monitoring system, wherein the monitoring system is preset with a plurality of preset warehouse identifiers, each of which corresponds to a warehouse to be monitored, each of which is installed with a plurality of monitoring cameras, and also includes a plurality of optional edge nodes, each of which is preset with a node recognition model, and different node recognition models can recognize different types of hazardous chemicals, and the method includes: Based on any of the preset warehouse identifiers, configure at least one target hazardous chemical type, determine the optional edge node corresponding to the target hazardous chemical type as an associated edge node, and record a mapping relationship between the preset warehouse identifier and the associated edge node in a preset node mapping table; Based on any of the associated edge nodes, creating respective associated candidate containers for the associated preset warehouse identifiers, and deploying the node identification model to the candidate containers; When any of the surveillance cameras is started and acquires a surveillance video stream, the target warehouse identifier of the associated warehouse to be monitored is determined, multiple target edge nodes are determined from the node mapping table based on the target warehouse identifier, and the target warehouse identifier is added to the surveillance video stream and then sent to each of the target edge nodes in parallel; Through any of the target edge nodes, each of the monitoring video streams is input into a corresponding target container based on the target warehouse identifier, and the corresponding target warehouse identifier is added to the model recognition result output by each of the target containers; Warehouse warning information is generated based on multiple model recognition results corresponding to the same target warehouse identifier.

[0006] According to some embodiments of the present invention, after determining a plurality of the target edge nodes from the node mapping table based on the target warehouse identifier, the method further includes: When the surveillance camera is started for the first time, a test video frame is acquired through the surveillance camera, and the corresponding preset warehouse identifier is determined as a test warehouse identifier; When at least one test hazard identifier is identified from the test video frame, a test hazard type corresponding to the test hazard identifier is determined, and the optional edge node corresponding to the test hazard type is determined as a test edge node; When the test edge node is associated with the test warehouse identifier in the node mapping table, acquiring the monitoring video stream through the monitoring camera; Alternatively, when the test edge node is not associated with the test warehouse identifier in the node mapping table, the candidate container corresponding to the test warehouse identifier is created through the test edge node, the association relationship between the test warehouse identifier and the test edge node is added to the node mapping table, and the monitoring video stream is obtained through the monitoring camera.

[0007] According to some embodiments of the present invention, the optional edge node is preset with multiple optional data sets, and the multiple optional data sets correspond to different security levels of the same node hazard type; after deploying the node identification model to the candidate container, the method further includes: Based on any of the candidate containers, obtaining a pre-associated warehouse security level based on the preset warehouse identifier, and determining a first data set from a plurality of the optional data sets based on the warehouse security level; Training the node recognition model based on the first data set; When a change in the warehouse security level is detected, the second data set is reacquired based on the changed warehouse security level, the current node recognition model is unloaded, and the new node recognition model is reloaded and trained based on the second data set.

[0008] According to some embodiments of the present invention, the plurality of surveillance cameras include at least one rotatable shared camera, the shared camera is pre-set with a reference video frame and a first direction, the shared camera is associated with a first warehouse and a second warehouse among the plurality of warehouses to be monitored, and the first warehouse is located in the first direction of the second warehouse; based on any of the shared cameras, when any of the surveillance cameras is started and a surveillance video stream is acquired, determining a target warehouse identifier of the associated warehouse to be monitored includes: After the shared camera is started, resetting the shared camera based on a first direction, acquiring a reference video stream, and rotating the shared camera based on a second direction, wherein the second direction is the opposite direction of the first direction; When the reference video frame is detected from the reference video stream, a first rotation duration is recorded, the shared camera is continuously controlled to rotate along the second direction to the other end, and a second rotation duration is recorded; resetting the shared camera based on the first direction again, determining the preset warehouse identifier of the first warehouse as the target warehouse identifier, and starting the shared camera to obtain the surveillance video stream; After controlling the shared camera to rotate based on the first rotation duration, determining the preset warehouse identifier of the second warehouse as the target warehouse identifier, and starting an identifier switching cycle; The identification switching cycle includes: When the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse, after the duration of continued rotation reaches twice the duration of the second rotation, the target warehouse identifier is switched to the preset warehouse identifier of the first warehouse, and after the duration of continued rotation reaches twice the duration of the first rotation, the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse.

[0009] According to some embodiments of the present invention, after deploying the node identification model to the candidate container, the method further includes: Select at least one standby warehouse identifier from the plurality of preset warehouse identifiers, and construct at least one standby time period based on each of the standby warehouse identifiers, wherein each of the standby time periods is associated with at least one standby hazardous chemical type; Based on any of the standby hazardous types, the corresponding target edge node is determined as a planned edge node, the associated standby warehouse identifier and the corresponding standby time period are used to construct monitoring plan information, and the monitoring plan information is sent to the planned edge node; Based on any of the planned edge nodes, the target container corresponding to the standby warehouse identifier is set to a standby state during the standby time period, and the communication of the monitoring camera connected to the target container is disconnected.

[0010] According to some embodiments of the present invention, after deploying the node identification model to the candidate container, the method further includes: Based on any of the target edge nodes, determine the communication resources of each of the surveillance cameras based on the current node resources, obtain the input video parameters preset by the node recognition model, and obtain the available video parameters of each of the surveillance cameras; Determine reference video parameters of each of the surveillance cameras based on the communication resources, the input video parameters, and the available video parameters, associate the reference video parameters with the target warehouse identifier of the corresponding surveillance camera, and send the reference video parameters to the corresponding surveillance camera, so that the surveillance camera obtains the surveillance video stream based on the reference video parameters; When the surveillance camera is the shared camera, each time the target warehouse identifier is switched, the reference video parameters associated with the switched target warehouse identifier are applied.

[0011] According to some embodiments of the present invention, after sending the reference video parameters to the corresponding surveillance camera, the method further includes: When any of the target edge nodes obtains a risk identification result, the monitoring camera corresponding to the monitoring video stream used to identify the risk identification result is determined as a target camera, wherein the risk identification result is the model identification result characterizing the risk of hazardous chemicals; Determine the optimal video parameters from the available video parameters of the target camera; when the reference video parameters applied by the target camera are different from the optimal video parameters, apply the optimal video parameters to continue to acquire the monitoring video stream; send a resource request and the optimal video parameters to the target edge node, so that the target edge node increases the communication resources with the target camera based on the optimal video parameters.

[0012] In a second aspect, an embodiment of the present invention provides a video surveillance device for a hazardous chemicals warehouse in a park, comprising at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the video surveillance method for the hazardous chemicals warehouse in a park as described in the first aspect above.

[0013] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a video surveillance device for a hazardous chemicals warehouse in a park as described in the second aspect above.

[0014] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the video surveillance method for a park hazardous chemicals warehouse as described in the first aspect above.

[0015] According to the video surveillance method of a hazardous chemical warehouse in a park according to an embodiment of the present invention, there are at least the following beneficial effects: based on any of the preset warehouse identifiers, at least one target hazardous chemical type is configured, the optional edge node corresponding to the target hazardous chemical type is determined as an associated edge node, and the mapping relationship between the preset warehouse identifier and the associated edge node is recorded in a preset node mapping table; based on any of the associated edge nodes, respective associated candidate containers are created for each associated preset warehouse identifier, and the node recognition model is deployed to the candidate container; when any of the monitoring cameras is started and a monitoring video stream is acquired, the target warehouse identifier of the associated warehouse to be monitored is determined, multiple target edge nodes are determined from the node mapping table based on the target warehouse identifier, the target warehouse identifier is added to the monitoring video stream, and then sent to each of the target edge nodes in parallel; through any of the target edge nodes, each of the monitoring video streams is input into the corresponding target container based on the target warehouse identifier, and the corresponding target warehouse identifier is added to the model recognition result output by each of the target containers; warehouse warning information is generated based on multiple model recognition results corresponding to the same target warehouse identifier. According to the technical solution of the embodiment of the present invention, candidate containers can be created for different warehouse identifiers at the edge node to configure the same recognition model, and the data isolation of different warehouses can be achieved by using the container; after the surveillance camera is configured with the target warehouse identifier, the surveillance video stream of the same surveillance camera is automatically sent to multiple edge nodes in parallel according to the node mapping table. Each edge node performs one type of recognition, which can achieve rapid recognition through a more lightweight recognition model, thereby improving the response efficiency and accuracy of hazardous chemicals monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1is a schematic diagram of a monitoring system provided by an embodiment of the present invention; Figure 2 is a flow chart of a video monitoring method for a park hazardous chemicals warehouse provided by another embodiment of the present invention; Figure 3 is a complete flow chart of a video monitoring method for a park hazardous chemicals warehouse provided by another embodiment of the present invention; Figure 4 It is a structural diagram of a video surveillance device for a park hazardous chemicals warehouse provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0017] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0018] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., and orientations or positional relationships indicated are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0019] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0020] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific content of the technical solution.

[0021] The embodiment of the present invention provides a video monitoring method, device, equipment, and storage medium for a hazardous chemical warehouse in a park, wherein the video monitoring method for the hazardous chemical warehouse in the park includes: configuring at least one target hazardous chemical type based on any of the preset warehouse identifiers, determining the optional edge node corresponding to the target hazardous chemical type as an associated edge node, and recording the mapping relationship between the preset warehouse identifier and the associated edge node in a preset node mapping table; creating respective associated candidate containers for each associated preset warehouse identifier based on any of the associated edge nodes, and deploying the node recognition model to the candidate container; when any of the monitoring cameras is started and obtains a monitoring video stream, determining the target warehouse identifier of the associated warehouse to be monitored, determining multiple target edge nodes from the node mapping table based on the target warehouse identifier, adding the target warehouse identifier to the monitoring video stream and sending them to each of the target edge nodes in parallel; through any of the target edge nodes, inputting each of the monitoring video streams into the corresponding target container based on the target warehouse identifier, and adding the corresponding target warehouse identifier to the model recognition results output by each of the target containers; generating warehouse warning information based on multiple model recognition results corresponding to the same target warehouse identifier. According to the technical solution of the embodiment of the present invention, candidate containers can be created for different warehouse identifiers at the edge node to configure the same recognition model, and the data isolation of different warehouses can be achieved by using the container; after the surveillance camera is configured with the target warehouse identifier, the surveillance video stream of the same surveillance camera is automatically sent to multiple edge nodes in parallel according to the node mapping table. Each edge node performs one type of recognition, which can achieve rapid recognition through a more lightweight recognition model, thereby improving the response efficiency and accuracy of hazardous chemicals monitoring.

[0022] First, refer to Figure 1 , Figure 1 A schematic diagram of a monitoring system provided for an embodiment of the present invention is applied to a monitoring system, wherein the monitoring system is preset with a plurality of preset warehouse identifiers, each preset warehouse identifier corresponds to a warehouse to be monitored, each warehouse to be monitored is installed with a plurality of monitoring cameras, and also includes a plurality of optional edge nodes, each optional edge node is preset with a node recognition model, and different node recognition models can recognize different types of hazardous chemicals.

[0023] It should be noted that the number of warehouses to be monitored in this embodiment is multiple, which is suitable for hazardous chemicals parks with multiple warehouses, such as Figure 1 As shown, the park includes a first warehouse 31 and a second warehouse 32. The hazardous chemicals in the first warehouse 31 and the second warehouse 32 can be the same or different. If they are the same, the same target hazardous chemical type can be recorded in the node mapping table. Each warehouse to be monitored can be equipped with multiple surveillance cameras, which can be installed indoors or outdoors in the warehouse to be monitored according to actual monitoring needs. For example Figure 1As shown, a first camera 311 is installed outdoors of the first warehouse 31, and abnormal personnel can be monitored through the first camera 311. A second camera 312 is installed indoors of the first warehouse 31, and inflammable and explosive hazardous chemicals placed inside are monitored through the second camera 312. A third camera 321 is installed indoors of the second warehouse 32, and the operating specifications of operators indoors are monitored through the third camera 321.

[0024] It should be noted that the monitoring system may be provided with a plurality of optional edge nodes, the optional edge nodes may be common edge servers, each optional edge node is provided with a node identification model, and the node identification models of different optional edge nodes may identify different contents, such as Figure 1 As shown, the node recognition model of the flammable and explosive identification node 21 is used to identify whether there are safety hazards such as metal collision or missing protective clothing in the flammable and explosive warehouse, the node recognition model of the abnormal personnel identification node 22 is used to detect abnormal personnel, the node recognition model of the solution splash identification node 23 is used to monitor whether there is solution splash during operation, and the node recognition model of the corrosive identification node 24 is used to monitor whether there are abnormalities in the stored corrosive hazardous chemicals. Specific node recognition models can be increased or decreased according to actual needs.

[0025] It should be noted that the node identification model of different hazardous types can be adjusted according to actual needs. This embodiment does not involve improvements in the model structure, identification principle and training method of the node identification model. Those skilled in the art can use existing models to complete relevant identification. Figure 1 In the flammable and explosive identification node 21 shown, the node identification model can adopt a 3D-CNN and voiceprint recognition fusion model to realize metal collision detection and missing protective clothing; for example, in the abnormal personnel identification node 22, HRNet key point detection and liquid morphology analysis can be used to detect whether the protective clothing is worn properly.

[0026] The following is based on the Figure 1 The monitoring system shown further illustrates the technical solution of the embodiment of the present invention.

[0027] Reference Figure 2 , Figure 2 A flowchart of a video monitoring method for a park hazardous chemicals warehouse provided by an embodiment of the present invention, the video monitoring method for a park hazardous chemicals warehouse includes but is not limited to the following steps: S10, based on any preset warehouse identifier, configure at least one target hazardous chemical type, determine the optional edge node corresponding to the target hazardous chemical type as an associated edge node, and record the mapping relationship between the preset warehouse identifier and the associated edge node in a preset node mapping table; S20, based on any associated edge node, creating respective associated candidate containers for each associated preset warehouse identifier, and deploying the node identification model to the candidate container; S30, when any surveillance camera is started and obtains the surveillance video stream, the target warehouse identifier of the associated warehouse to be monitored is determined, multiple target edge nodes are determined from the node mapping table based on the target warehouse identifier, and the target warehouse identifier is added to the surveillance video stream and sent to each target edge node in parallel; S40, through any target edge node, input each monitoring video stream into the corresponding target container based on the target warehouse identifier, and add the corresponding target warehouse identifier to the model recognition result output by each target container; S50, generating warehouse warning information based on multiple model recognition results corresponding to the same target warehouse identifier.

[0028] It should be noted that warehouses within the hazardous chemicals park can be distinguished by different preset warehouse identifiers, which can uniquely identify different warehouses to be monitored. Since hazardous chemicals belong to an industry with extremely high safety requirements, this embodiment can generate a visual configuration interface in the monitoring system. Professional managers can select any preset warehouse identifier in the configuration interface, and then display all optional hazardous chemical types in the configuration interface. Then, the target hazardous chemical type can be selected from the optional hazardous chemical types according to the actual needs of the warehouse to be monitored, so as to avoid safety risks caused by errors in the automatic allocation of hazardous chemical types. The optional hazardous chemical types and target hazardous chemical types can be Figure 1 The flammable and explosive, abnormal personnel, solution splashing, corrosiveness, etc. shown in are used to indicate the monitoring content of hazardous chemicals. The number and types of specific target hazardous chemicals are not limited here.

[0029] It should be noted that each optional edge node is uniquely associated with a node hazardous chemical type, so the corresponding associated edge node can be matched according to the target hazardous chemical type selected in the monitoring system, and the safety requirements of each hazardous chemical warehouse are known, and a warehouse can include multiple types of target hazardous chemical types, so each preset warehouse identifier can have multiple associated edge nodes, and each node mapping table is constructed based on this, which serves as the information basis for subsequent configuration of surveillance cameras.

[0030] For example, Figure 1As shown, the preset warehouse identifier of the first warehouse 31 is warehouse identifier A, and the preset warehouse identifier of the second warehouse 32 is warehouse identifier B. In the monitoring system, two target hazardous chemical types, namely flammable and explosive and abnormal personnel, are specified for warehouse identifier A, and three target hazardous chemical types, namely abnormal personnel, solution splash and corrosive, are specified for warehouse identifier B. The mapping relationship between warehouse identifier A and flammable and explosive identification node 21 and abnormal personnel identification node 22 is recorded in the node mapping table, and the mapping relationship between warehouse identifier B and abnormal personnel identification node 22, solution splash identification node 23 and corrosive identification node 24 is recorded.

[0031] It should be noted that in order to achieve full coverage of safety monitoring, the monitoring system can configure optional associated nodes for all optional hazardous chemical types. If the actual warehouse does not involve the relevant hazardous chemical types, for example, all warehouses to be monitored do not involve the safety requirements of protecting flammable and explosive materials, then the corresponding optional edge node will not be determined as an associated edge node, and the optional edge node will not perform subsequent operations. This embodiment only discusses the subsequent process for the node determined as an associated edge node.

[0032] It should be noted that in the associated edge node, this embodiment uses container technology to load the model, and constructs candidate containers according to the number of preset warehouse identifiers associated with the associated edge node. Each warehouse corresponds to a candidate container, and the preset warehouse identifier is bound to the candidate container. When multiple surveillance cameras send video streams to the same associated edge node at the same time, the preset warehouse identifier is used as the basis for identifying whether the video stream can be input into the candidate container. In addition, container technology can achieve data isolation, avoid misidentification caused by confusing video streams from different warehouses, and improve the reliability of recognition results. Multiple candidate containers of the same associated edge node deploy the same node recognition model, so that intelligent recognition of the same hazardous chemical type can be achieved in each candidate container.

[0033] For example, Figure 1 As shown, the first warehouse 31 is associated with the inflammable and explosive identification node 21 and the abnormal personnel identification node 22. A candidate container is deployed in the inflammable and explosive identification node 21, and an inflammable and explosive identification model is deployed so that the video stream of the first camera 311 can be input into the inflammable and explosive identification model for related identification. The first container 221 is deployed for the first warehouse 31 in the abnormal personnel identification node 22. Since the second warehouse 32 is also associated with the abnormal personnel identification node 22, the second container 222 is deployed for the second warehouse 32 in the abnormal personnel identification node 22. The same abnormal personnel identification model is deployed in the first container 221 and the second container 222. The video stream of the first camera 311 is input into the first container 211 for identification, and the video stream of the third camera 321 is input into the second container 222 for identification.

[0034] It should be noted that before the surveillance camera is started, the warehouse to be monitored to which each surveillance camera belongs is known and fixed. Therefore, the target warehouse identifier of each surveillance camera can be pre-configured in the monitoring system. The surveillance camera can send the same surveillance video stream to each target edge node in parallel based on the target edge node associated with the target warehouse identifier, and record the target warehouse identifier in the surveillance video stream, so that the target edge node can determine the target container that needs to be input into the surveillance video stream according to the target warehouse identifier, and identify it to obtain the model recognition result.

[0035] In this embodiment, the monitoring system can send the communication address of the target edge node to the monitoring camera according to the node mapping table. The monitoring camera establishes connections based on multiple communication addresses to obtain multiple channels, and sends the monitoring video stream to each channel, so that each target edge node can obtain the monitoring video stream. Of course, in the case where the monitoring camera does not have processing capabilities, the monitoring video stream can also be sent to the monitoring system in its entirety, and the monitoring system will synchronously distribute it to each target edge node according to the node mapping table. The forwarding method of the monitoring video stream can be selected according to the actual hardware situation, and it can ensure that the target edge node obtains the correct monitoring video stream.

[0036] It is worth noting that this embodiment sends the same monitoring video stream to multiple target edge nodes for identification. Although it takes up more resources in data transmission compared to the solution of identification by one recognition model, in the field of hazardous chemicals monitoring, response efficiency is far more important than resource utilization. This embodiment uses different target edge nodes to perform different types of identification, such as Figure 1 As shown, the video stream of the first camera 311 is simultaneously input to the inflammable and explosive identification node 21 and the abnormal person identification node 22, respectively. Inflammable and explosive identification is performed through the inflammable and explosive identification node 21, and abnormal person identification is performed through the abnormal person identification node 22. Compared with simultaneously performing inflammable and explosive identification and abnormal person identification based on the same video stream, the calculation amount of a single model in this embodiment is significantly smaller than that of the related technology solutions, and the model identification results can be output faster, thereby improving the response efficiency of hazardous chemicals monitoring.

[0037] It should be noted that after obtaining the model recognition result, the model recognition result is fed back to the monitoring system after adding the target warehouse identification. The target warehouse identification is the basis for the fusion of multiple model recognition results. If the model recognition result does not indicate an abnormality, the monitoring system does not need to generate warehouse warning information. If an abnormality occurs, the warehouse warning information is generated based on all model results of the target warehouse identification with the abnormality. The warehouse warning information only needs to describe the abnormal situation. Figure 1As shown, when an abnormal person is detected in the first warehouse 31, the model recognition result output by the abnormal person recognition node 22 is "abnormal person detected", while the flammable and explosive identification node 21 does not output the model recognition result representing the abnormality. Since both carry the warehouse identification A, the two model recognition results can be fused based on the warehouse identification A, and the acoustic detection will give you the warehouse warning information of "abnormal person detected".

[0038] In addition, in one embodiment, referring to Figure 3 In step S30, after determining multiple target edge nodes from the node mapping table based on the target warehouse identifier, the following steps are also included but not limited to: S311, when the surveillance camera is started for the first time, a test video frame is obtained through the surveillance camera, and the corresponding preset warehouse identifier is determined as the test warehouse identifier; S312, when at least one test hazard identifier is identified from the test video frame, a test hazard type corresponding to the test hazard identifier is determined, and an optional edge node corresponding to the test hazard type is determined as a test edge node; S313, when the test edge node is associated with the test warehouse identifier in the node mapping table, a monitoring video stream is obtained through a monitoring camera; S314, when the test edge node is not associated with the test warehouse identifier in the node mapping table, a candidate container corresponding to the test warehouse identifier is created through the test edge node, an association relationship between the test warehouse identifier and the test edge node is added to the node mapping table, and a monitoring video stream is obtained through a monitoring camera.

[0039] It should be noted that, according to the description of the above embodiment, each target edge node corresponds to a target hazardous chemical type, that is, it can realize the risk identification of a class of hazardous chemicals. The node mapping table is obtained based on the configuration operation of the management personnel, and the management personnel may also have the possibility of configuration errors. Since the hazardous chemical warehouse must be coated with hazardous chemical type identification, such as flammable and explosive identification, corrosive gas identification, etc., these identifications are standard patterns and are usually within the shooting range of at least one surveillance camera. Therefore, this embodiment obtains a test video frame after each surveillance camera is started for the first time, and uses all known hazardous chemical identifications to perform object recognition on the test video frame, and determines the identified as a test hazardous chemical identification. Since the hazardous chemical type corresponding to the test hazardous chemical identification is knowable and unique, this embodiment uses this as the test hazardous chemical type to determine the test edge node corresponding to the test hazardous chemical type. If the test edge node exists in the node mapping table and is associated with the test warehouse identification, it can be determined that the configuration of the node mapping table is correct and no hazardous chemical type is missed. If the test edge node is not associated with the test warehouse identifier in the node mapping table, it can be determined that the test hazardous chemical type is omitted during the configuration process. According to the steps of the above embodiment, a candidate container is automatically created in the test edge node, associated with the test warehouse identifier and recorded in the node mapping table. The technical solution of this embodiment can verify and automatically correct the manual configuration to ensure that the node mapping table records all hazardous chemical types that need to be monitored in the monitored warehouse.

[0040] For example, Figure 1 As shown, taking the third camera 321 as an example, the test video frame shot after the third camera 321 is started records the corrosive mark, the abnormal person mark and the solution splash mark. If the node mapping table only records the abnormal person identification node 22 and the solution splash identification node 23, then the candidate container is constructed at the corrosive identification node 24 and associated with the warehouse identification B, and the corrosive identification node 24 is added to the corresponding item of the warehouse identification B in the node mapping table.

[0041] In addition, in one embodiment, referring to Figure 3 The optional edge node is preset with multiple optional data sets, and the multiple optional data sets correspond to different security levels of the same node hazard type; after executing step S20, it also includes but is not limited to the following steps: S21, based on any candidate container, obtaining a pre-associated warehouse security level based on a preset warehouse identifier, and determining a first data set from a plurality of optional data sets based on the warehouse security level; S22, training a node recognition model based on the first data set; S23, when a change in the warehouse safety level is detected, the second data set is reacquired based on the changed warehouse safety level, the current node recognition model is unloaded, and the new node recognition model is reloaded and trained based on the second data set.

[0042] It should be noted that, according to the description of the above embodiment, each optional edge node corresponds to the same node recognition model, and can be identified based on the video streams of surveillance cameras in different warehouses. Different warehouses to be monitored may have different safety standards for hazardous chemicals. Since the optional edge node not only has processing capabilities but also has certain storage capabilities, this embodiment configures multiple optional data sets in the optional edge node, and each optional data set corresponds to a security level. After constructing the node mapping table, the node recognition model is loaded into the candidate container. This embodiment further obtains the warehouse safety level according to a preset warehouse identifier, selects a first data set from multiple optional data sets according to the warehouse safety level, and trains the node recognition model in the candidate container through the first data set, so that the node recognition model can be identified according to the safety level requirements of the warehouse to be monitored.

[0043] For example, the requirement of the first warehouse 31 is that any employee will not be identified as an abnormal person, while the requirement of the second warehouse 32 is that only employees with specific permissions will not be identified as abnormal persons. Or the requirement of the first warehouse 31 for inflammable and explosive materials is only that there should be no open flames, while the requirement of the second warehouse 32 for inflammable and explosive materials is that there should be no static electricity, which is higher than the requirement of the first warehouse 31. Taking abnormal personnel as an example, the abnormal personnel identification node 22 is provided with a high security level data set, a medium security level data set and a low security level data set. The first data set determined by the first warehouse 31 is a medium security level data set, and the first data set determined by the second warehouse 32 is a high security level data set. Model training is performed in the first container 221 based on the medium security level data set, and model training is performed in the second container 222 based on the high security level data set, so that after the first container 221 obtains the monitoring video stream, it can output the risk identification result after identifying that the person is not an employee, and after the second container 222 obtains the monitoring video stream, it outputs the risk identification result after identifying that the person is not a specific employee.

[0044] In addition, in one embodiment, referring to Figure 3 The multiple surveillance cameras include at least one rotatable shared camera, the shared camera is pre-set with a reference video frame and a first direction, the shared camera is associated with a first warehouse and a second warehouse among the multiple warehouses to be monitored, and the first warehouse is located in the first direction of the second warehouse; based on any shared camera, in step S30, specifically including but not limited to the following steps: S321, after the shared camera is started, resetting the shared camera based on the first direction, acquiring a reference video stream, and rotating the shared camera based on a second direction, wherein the second direction is the opposite direction of the first direction; S322, when a reference video frame is detected from the reference video stream, a first rotation duration is recorded, and the shared camera is continuously controlled to rotate along a second direction to the other end, and a second rotation duration is recorded; S323, resetting the shared camera based on the first direction again, determining the preset warehouse identifier of the first warehouse as the target warehouse identifier, and starting the shared camera to obtain a surveillance video stream; S324, after controlling the shared camera to rotate based on the first rotation duration, determining the preset warehouse identifier of the second warehouse as the target warehouse identifier, and starting an identifier switching cycle; The logo switching cycle includes: When the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse, after the duration of continued rotation reaches twice the duration of the second rotation, the target warehouse identifier is switched to the preset warehouse identifier of the first warehouse, and after the duration of continued rotation reaches twice the duration of the first rotation, the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse.

[0045] It should be noted that the geographical scope of the hazardous chemicals park is usually large. The surveillance camera can be set not only on the wall of the warehouse to be monitored, but also on the installable body outside the warehouse to be monitored, such as street lights, electric poles, etc. In this case, the surveillance camera cooperates with the steering structure to realize scanning monitoring, such as for monitoring Figure 1 Whether there are any abnormal persons in the open space between the first warehouse 31 and the second warehouse 32. The rotating structure of the monitoring camera is a technology well known to those skilled in the art, and will not be described in detail here.

[0046] It should be noted that, according to the description of the above embodiment, different warehouses to be monitored correspond to different target edge nodes. Since the shared camera includes two warehouses to be monitored, it corresponds to two groups of target edge nodes. In order to ensure that the monitoring video stream is transmitted to the correct target edge node, this embodiment pre-sets a reference video frame. For example, after completing the installation of the monitoring camera, the installer can aim the monitoring camera at the middle area of ​​the two warehouses to be monitored and then start it. A frame of the current video stream can be captured as a reference video frame, and the switching position between the two warehouses to be monitored can be represented by the reference video frame.

[0047] It should be noted that the first direction can be adjusted according to the actual positions of the first warehouse and the second warehouse. For example, if the first warehouse is located on the left side of the second warehouse, the first direction is the left side and the second direction is the right side. After the shared camera is started, it is reset to one end of the first direction, moves to the other end and obtains the reference video stream. When the reference video frame is detected, it is determined that the position has been reached to switch from monitoring the first warehouse to monitoring the second warehouse. The duration of rotation is determined as the first rotation duration, and the camera continues to rotate in the second direction. After rotating to the farthest end of the second direction, the second rotation duration is recorded. Since the rotation speed of the surveillance camera is uniform, the switching timing can be characterized by the first rotation duration and the second rotation duration.

[0048] It should be noted that after resetting the shared camera and starting it, the monitoring object of the shared camera is the first warehouse, so its preset warehouse identifier is determined as the target warehouse identifier, and the monitoring video stream is sent to the target edge node corresponding to the first warehouse. When the first rotation duration is the first rotation duration, the switching point is reached, and the target warehouse identifier is automatically replaced with the preset warehouse identifier of the second warehouse, and the monitoring video stream is sent to the target edge node of the second warehouse. After completing the first switch, without considering the pause of the endpoint, the time it takes for the shared camera to turn back to the switching point is twice the second rotation duration. After switching to the first warehouse, the time it takes to return to the switching point is twice the first rotation duration. This embodiment switches the target warehouse identifier based on this time duration to ensure that the monitoring video stream is transmitted to the correct target edge node. Of course, if the pause duration of the endpoint is considered, the pause duration can also be added after twice the first rotation duration, which will not be repeated here.

[0049] In addition, in one embodiment, referring to Figure 3 After executing step S20, the following steps are also included but not limited to: S24, randomly selecting at least one standby warehouse identifier from a plurality of preset warehouse identifiers, and constructing at least one standby time period based on each standby warehouse identifier, wherein each standby time period is associated with at least one standby hazardous chemical type; S25, based on any standby hazardous type, determine the corresponding target edge node as a planned edge node, construct monitoring plan information with the associated standby warehouse identifier and the corresponding standby time period, and send the monitoring plan information to the planned edge node; S26, based on any planned edge node, setting the target container corresponding to the standby warehouse identifier to a standby state during the standby period, and disconnecting the communication of the surveillance camera connected to the target container.

[0050] It should be noted that in the field of hazardous chemicals monitoring, continuous monitoring is required in some scenarios, such as abnormal personnel monitoring, flammable and explosive monitoring, etc., while there are also some scenarios that do not require continuous monitoring, such as solution splash monitoring, which will only occur when there are operators operating. After the operator gets off work, there is no hazardous chemical operation, so there will be no solution splash. In the above embodiment, the monitoring camera will send the monitoring video stream to each target edge node in parallel. In order to reduce unnecessary waste of resources, this embodiment configures the connection status of the planned edge node and the monitoring camera through the standby warehouse identifier and the standby time period, and disconnects the communication of the monitoring camera associated with the standby warehouse identifier during the standby time period, and no longer receives the monitoring video stream. Of course, the planned edge node can be connected to multiple monitoring cameras. If the corresponding preset warehouse identifier is not configured as a standby warehouse identifier, the planned edge node still receives the relevant monitoring video stream and performs monitoring operations. This embodiment does not start and stop the planned edge node according to the standby time period, but only stops receiving the monitoring video stream of the warehouse to be monitored that does not need to be monitored.

[0051] It should be noted that during the configuration process, one or more standby warehouse identifiers can be selected in the monitoring system, and the various standby hazardous chemical types that have been bound can be displayed in the visual interface. The management personnel can configure a standby time period for the standby warehouse identifier, and configure the standby hazardous chemical type for the standby time point. The monitoring system determines the planned edge node according to the standby hazardous chemical type, and sends the monitoring plan information to the planned edge node, so that the planned edge node determines the corresponding surveillance camera based on the standby warehouse identifier during the standby time period, disconnects the corresponding surveillance camera to stop receiving the video stream, and the corresponding target container has no further input, and the node recognition model no longer has input data and enters standby, which not only saves network transmission resources, but also saves the computing resources of the planned edge node.

[0052] For example, Figure 1 As shown, taking the abnormal personnel identification node as an example, when the standby time period of the abnormal personnel identification node 22 is configured from 20:00 to 8:00 the next day based on the warehouse identification A, the abnormal personnel identification node disconnects from the first camera 311 and the second camera 312 during this time period, and maintains the connection with the third camera 321. No identification operation is performed in the first container 221, and the identification of the surveillance video stream is maintained in the second container 222.

[0053] In addition, in one embodiment, referring to Figure 3 After executing step S20, the following steps are also included but not limited to: S27, based on any target edge node, determine the communication resources of each surveillance camera based on the current node resources, obtain the input video parameters preset by the node recognition model, and obtain the available video parameters of each surveillance camera; S28, determining reference video parameters of each surveillance camera based on the communication resources, the input video parameters, and the available video parameters, associating the reference video parameters with the target warehouse identifier of the corresponding surveillance camera, and sending the reference video parameters to the corresponding surveillance camera, so that the surveillance camera obtains a surveillance video stream based on the reference video parameters; S29, when the surveillance camera is a shared camera, each time the target warehouse identifier is switched, the reference video parameters associated with the switched target warehouse identifier are applied.

[0054] It should be noted that the specifications of surveillance cameras may be different. For example, some models whose recognition processes rely on subtle images require input of higher-resolution video streams, while some simple target detection models only require input of lower-resolution video streams. In order to improve the utilization of transmission resources and computing resources of edge nodes, this embodiment determines the communication resources of each associated surveillance camera based on node resources through the target edge node, such as determining the available bandwidth resources for each surveillance camera. Since the node recognition models are the same, it is only necessary to obtain the input video parameters once, and determine the reference video parameters in combination with the available video parameters of the surveillance camera. For example, Figure 1 As shown, the first camera 311 supports 4k60 frame shooting, the third camera 321 only supports 720p30 frame shooting, and the input video parameter required by the node recognition model is 720p. In order to save transmission resources and computing resources, the same bandwidth resources are allocated to the first camera 311 and the third camera 321 based on 720p, and the first camera 311 and the third camera 321 are notified to obtain the monitoring video stream with a resolution of 720p. The shooting parameters of the monitoring camera can be dynamically adjusted according to the parameters of the monitoring camera, the node recognition model and the actual needs of the target edge node.

[0055] It should be noted that when there are multiple target edge nodes, it is necessary to ensure that the reference video parameters can meet the requirements of all target edge nodes, for example Figure 1 As shown, the flammable and explosive identification node 21 requires at least a 1080p resolution, while the abnormal person identification node 22 requires at least a 720p resolution, and the reference video parameter is 1080p which is upward compatible.

[0056] It should be noted that when the surveillance camera is a shared camera, the reference video parameters can be switched according to the above-mentioned switching time point, which will not be elaborated here.

[0057] In addition, in one embodiment, referring to Figure 3 After executing step S28, the following steps are also included but not limited to: S281, when any target edge node obtains a risk identification result, the surveillance camera corresponding to the surveillance video stream used to identify the risk identification result is determined as the target camera, wherein the risk identification result is a model identification result characterizing the risk of hazardous chemicals; S282, determine the optimal video parameters from the available video parameters of the target camera. When the reference video parameters applied by the target camera are different from the optimal video parameters, apply the optimal video parameters to continue to obtain the monitoring video stream, and send a resource request and the optimal video parameters to the target edge node, so that the target edge node increases the communication resources with the target camera based on the optimal video parameters.

[0058] It should be noted that according to the description of the above embodiment, the reference video parameters can be adjusted dynamically. When the target edge node obtains the risk identification result, there may be safety risks of hazardous chemicals in the current monitoring range. In order to avoid misidentification or to accurately identify further risks, this embodiment adjusts the target camera to the optimal video parameters and transmits the highest definition video stream to the target edge node, thereby improving the identification accuracy when risks occur.

[0059] Exemplarily, taking the third camera 321 as an example, the current reference video parameter is 720p. When the risk identification result of solution splash is detected, the optimal video parameter of 4k60 frames is applied to transmit high-definition and high-frame rate monitoring video stream to the solution splash identification node, which can more accurately identify whether solution splash continues to occur and improve the accuracy of early warning.

[0060] like Figure 4 As shown, Figure 4 The present invention also provides a video monitoring device for a park hazardous chemicals warehouse, including: The processor 401 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application; The memory 402 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 402 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 402, and the processor 401 calls and executes the video monitoring method for the park hazardous chemicals warehouse in the embodiment of this application; Input / output interface 403, used to implement information input and output; Communication interface 404, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.); A bus 405 that transmits information between various components of the device (e.g., processor 401, memory 402, input / output interface 403, and communication interface 404); The processor 401 , the memory 402 , the input / output interface 403 and the communication interface 404 are connected to each other in communication within the device via the bus 405 .

[0061] An embodiment of the present application also provides an electronic device, including the video surveillance device for a park hazardous chemicals warehouse as described above.

[0062] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the video monitoring method for the hazardous chemicals warehouse in the park is implemented.

[0063] As a non-transient computer-readable storage medium, the memory can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and are implemented to be located in one place, or may also be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.

[0064] It will be appreciated by those skilled in the art that all or some of the steps and systems in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically include computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0065] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions under the shared conditions without violating the spirit of the present invention. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A video monitoring method for a hazardous chemicals warehouse in a park, characterized in that: Applied to a monitoring system, the monitoring system is preset with a plurality of preset warehouse identifiers, each of the preset warehouse identifiers corresponds to a warehouse to be monitored, each of the warehouses to be monitored is installed with a plurality of monitoring cameras, and also includes a plurality of optional edge nodes, each of the optional edge nodes is preset with a node recognition model, and different node recognition models can recognize different hazardous chemical types, the method includes: Based on any of the preset warehouse identifiers, configure at least one target hazardous chemical type, determine the optional edge node corresponding to the target hazardous chemical type as an associated edge node, and record a mapping relationship between the preset warehouse identifier and the associated edge node in a preset node mapping table; Based on any of the associated edge nodes, creating respective associated candidate containers for the associated preset warehouse identifiers, and deploying the node identification model to the candidate containers; When any of the surveillance cameras is started and acquires a surveillance video stream, the target warehouse identifier of the associated warehouse to be monitored is determined, multiple target edge nodes are determined from the node mapping table based on the target warehouse identifier, and the target warehouse identifier is added to the surveillance video stream and then sent to each of the target edge nodes in parallel; Through any of the target edge nodes, each of the monitoring video streams is input into a corresponding target container based on the target warehouse identifier, and the corresponding target warehouse identifier is added to the model recognition result output by each of the target containers; Warehouse warning information is generated based on multiple model recognition results corresponding to the same target warehouse identifier.

2. The video monitoring method for a park hazardous chemicals warehouse according to claim 1 is characterized in that: After determining a plurality of the target edge nodes from the node mapping table based on the target warehouse identifier, the method further includes: When the surveillance camera is started for the first time, a test video frame is acquired through the surveillance camera, and the corresponding preset warehouse identifier is determined as a test warehouse identifier; When at least one test hazard identifier is identified from the test video frame, a test hazard type corresponding to the test hazard identifier is determined, and the optional edge node corresponding to the test hazard type is determined as a test edge node; When the test edge node is associated with the test warehouse identifier in the node mapping table, acquiring the monitoring video stream through the monitoring camera; Alternatively, when the test edge node is not associated with the test warehouse identifier in the node mapping table, the candidate container corresponding to the test warehouse identifier is created through the test edge node, the association relationship between the test warehouse identifier and the test edge node is added to the node mapping table, and the monitoring video stream is obtained through the monitoring camera.

3. The video monitoring method for a park hazardous chemicals warehouse according to claim 1 is characterized in that: The optional edge node is preset with a plurality of optional data sets, and the plurality of optional data sets correspond to different security levels of the same node hazard type; After deploying the node identification model to the candidate container, the method further includes: Based on any of the candidate containers, obtaining a pre-associated warehouse security level based on the preset warehouse identifier, and determining a first data set from a plurality of the optional data sets based on the warehouse security level; Training the node recognition model based on the first data set; When a change in the warehouse security level is detected, the second data set is reacquired based on the changed warehouse security level, the current node recognition model is unloaded, and the new node recognition model is reloaded and trained based on the second data set.

4. The video monitoring method for a park hazardous chemicals warehouse according to claim 1 is characterized in that: The plurality of monitoring cameras include at least one rotatable shared camera, the shared camera is pre-set with a reference video frame and a first direction, the shared camera is associated with a first warehouse and a second warehouse among the plurality of warehouses to be monitored, and the first warehouse is located in the first direction of the second warehouse; Based on any of the shared cameras, when any of the surveillance cameras is started and a surveillance video stream is acquired, determining a target warehouse identifier of the associated warehouse to be monitored includes: After the shared camera is started, resetting the shared camera based on a first direction, acquiring a reference video stream, and rotating the shared camera based on a second direction, wherein the second direction is the opposite direction of the first direction; When the reference video frame is detected from the reference video stream, a first rotation duration is recorded, the shared camera is continuously controlled to rotate along the second direction to the other end, and a second rotation duration is recorded; resetting the shared camera based on the first direction again, determining the preset warehouse identifier of the first warehouse as the target warehouse identifier, and starting the shared camera to obtain the surveillance video stream; After controlling the shared camera to rotate based on the first rotation duration, determining the preset warehouse identifier of the second warehouse as the target warehouse identifier, and starting an identifier switching cycle; The identification switching cycle includes: When the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse, after the duration of continued rotation reaches twice the duration of the second rotation, the target warehouse identifier is switched to the preset warehouse identifier of the first warehouse, and after the duration of continued rotation reaches twice the duration of the first rotation, the target warehouse identifier is switched to the preset warehouse identifier of the second warehouse.

5. The video monitoring method for a park hazardous chemicals warehouse according to claim 4 is characterized in that: After deploying the node identification model to the candidate container, the method further includes: Select at least one standby warehouse identifier from the plurality of preset warehouse identifiers, and construct at least one standby time period based on each of the standby warehouse identifiers, wherein each of the standby time periods is associated with at least one standby hazardous chemical type; Based on any of the standby hazardous types, the corresponding target edge node is determined as a planned edge node, the associated standby warehouse identifier and the corresponding standby time period are used to construct monitoring plan information, and the monitoring plan information is sent to the planned edge node; Based on any of the planned edge nodes, the target container corresponding to the standby warehouse identifier is set to a standby state during the standby time period, and the communication of the monitoring camera connected to the target container is disconnected.

6. The video monitoring method for a park hazardous chemicals warehouse according to claim 5 is characterized in that: After deploying the node identification model to the candidate container, the method further includes: Based on any of the target edge nodes, determine the communication resources of each of the surveillance cameras based on the current node resources, obtain the input video parameters preset by the node recognition model, and obtain the available video parameters of each of the surveillance cameras; Determine reference video parameters of each of the surveillance cameras based on the communication resources, the input video parameters, and the available video parameters, associate the reference video parameters with the target warehouse identifier of the corresponding surveillance camera, and send the reference video parameters to the corresponding surveillance camera, so that the surveillance camera obtains the surveillance video stream based on the reference video parameters; When the surveillance camera is the shared camera, each time the target warehouse identifier is switched, the reference video parameters associated with the switched target warehouse identifier are applied.

7. The video monitoring method for a park hazardous chemicals warehouse according to claim 6 is characterized in that: After sending the reference video parameters to the corresponding surveillance camera, the method further includes: When any of the target edge nodes obtains a risk identification result, the monitoring camera corresponding to the monitoring video stream used to identify the risk identification result is determined as a target camera, wherein the risk identification result is the model identification result characterizing the risk of hazardous chemicals; Determine the optimal video parameters from the available video parameters of the target camera; when the reference video parameters applied by the target camera are different from the optimal video parameters, apply the optimal video parameters to continue to acquire the monitoring video stream; send a resource request and the optimal video parameters to the target edge node, so that the target edge node increases the communication resources with the target camera based on the optimal video parameters.

8. A video monitoring device for a hazardous chemicals warehouse in a park, characterized in that: It includes at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the video surveillance method for a park hazardous chemicals warehouse as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: Including the video monitoring device for the park hazardous chemicals warehouse as described in claim 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the video monitoring method for a park hazardous chemicals warehouse as described in any one of claims 1 to 7.

Images