Reservoir Internet of Things monitoring alarm method, system and equipment based on YOLOv5 algorithm

Abstract

The invention provides a reservoir Internet of Things monitoring alarm method, system and equipment based on a YOLOv5 algorithm. According to the system, a panoramic camera is deployed on a floating platform, the YOLOv5 algorithm is utilized to carry out deep analysis on collected images, the position and feature information of a human body and other targets is accurately extracted, and a normal state, an early warning state or an alarm state are judged and classified. When an emergency situation is detected, not only is alarm information generated and sent to a manager, but also intervention means can be automatically started, and the intervention means comprise automatic deployment of lifesaving equipment, system triggering of release of an automatic life buoy, rapid reaching of a designated position and providing of instant rescue support for a drowning person. The system simulates a real scene in a water area through training of a big data analysis model, so that private underwater drowning events caused by unmanned or insufficient manual management are remarkably reduced, and the safety monitoring level of the reservoir is comprehensively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of safety monitoring, and in particular relates to a reservoir Internet of Things monitoring alarm method, system and equipment based on the YOLOv5 algorithm. Background Art

[0002] Drowning incidents often occur in reservoirs, rivers and other waters. In these areas, the water conditions are complex and there are few people. Even if a drowning incident occurs, it is difficult to rescue in time. Therefore, it is an urgent problem to monitor these waters and prohibit people from entering the water privately.

[0003] At present, it is difficult to equip reservoirs and other environments with professional managers, and the river basin is wide and large, so a single manager cannot monitor and manage the entire river basin around the clock. Most cameras in reservoirs are not equipped with intelligent detection models to monitor illegal behaviors such as private entry into the water. The traditional YOLO model has large errors in real-time detection of people entering the water and preventing drowning. The targets on the water surface may be partially or completely blocked by other objects, resulting in a reduction in detectable feature information. Only part of the target may be exposed above the water surface, making the complete information obtained from the water surface insufficient. Some targets, especially people or small objects in the distance, may only occupy a small area in the image and are difficult to be accurately identified. These factors together increase the difficulty of target detection in water scenes, and the accuracy and robustness of detection need to be further improved. In addition, the existing detection methods do not effectively divide areas and dangerous situations, which will also lead to a large number of false detections. More importantly, most detection models only have detection functions, and further rescue measures still need to be provided by managers. In the absence of managers, only detection without intervention is of little significance for drowning incidents. Summary of the invention

[0004] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide a reservoir Internet of Things monitoring and alarm method, system and equipment based on the YOLOv5 algorithm. The first purpose is to solve the problem of false detection caused by undivided areas. The second purpose is to solve the problem that the existing model is difficult to accurately identify small targets. The third purpose is to solve the problem that the prior art only checks without life-saving means, thereby comprehensively increasing the safety factor of reservoirs and other basins.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A first aspect of the present invention provides a reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm, comprising the following steps:

[0007] When it is determined that someone has entered the water monitoring range, the visual sensing technology is used to continuously capture images of the water area including the edge of the reservoir;

[0008] After resizing and normalizing the water area image, the YOLOv5 algorithm is used to obtain the personnel location information, and the current state type of the water area is determined according to the location information, including normal state, warning state and alarm state; the normal state means that there are no personnel in the reservoir and at the edge of the reservoir, the warning state means that there are personnel at the edge of the reservoir but no personnel in the reservoir, and the alarm state means that there are personnel in the reservoir;

[0009] Generate alarm information based on the status type and send it to the management terminal through the Internet of Things;

[0010] Among them, when it is in the alarm state, the surface lifesaving equipment is controlled to move to the personnel position coordinates.

[0011] In one embodiment, the water area monitoring range is 5 meters away from the edge of the reservoir surface, and the edge of the reservoir is 2 meters away from the edge of the reservoir surface. Personnel information is collected through a camera, and the YOLOv5 algorithm is used to locate the personnel and analyze the distance between them and the edge of the reservoir. When the distance is less than 2 meters, it can be determined that someone has entered the edge of the reservoir. At the same time, infrared sensors are installed on the water surface life-saving equipment, combined with the detection results of the image sensor, for dual protection, and real-time tracking and warning of personnel status.

[0012] In one embodiment, the YOLOv5 algorithm is used to obtain the personnel location information, and the implementation method is as follows:

[0013] First, the preprocessed water image is sliced ​​through the fcous layer to reduce the amount of convolution calculation. Then YOLOv5 uses the backbone network architecture to extract the features of the human body in the water image, including shape, spatial relationship, scale, context and motion. The obtained features enter the neck module and are fused through the feature pyramid network and path aggregation network; the fused features enter the Head module for final target prediction; in the backbone network architecture, the CSP module is replaced by a module containing CBAM or ECA.

[0014] In one embodiment, in order to more accurately extract the characteristics of the human body in the water, the present invention also combines the YOLOv5 algorithm with the MediaPipe model, uses the MediaPipe model to obtain the key point information of the human body, estimates the posture of the human body, and determines whether drowning occurs. For example, the drowning characteristics such as lowering the head, raising the head, opening the mouth, closing the eyes, etc., take the mouth as an example, according to the key point information of the mouth, calculate the length and width of the mouth, so as to determine whether the mouth is open, the number of times the mouth is opened, etc. If an abnormality is found, the alarm device is triggered in time.

[0015] In one embodiment, the generation of alarm information according to the status type is implemented as follows:

[0016] When the current status is normal, remain silent and continue monitoring;

[0017] When the current status is the warning status, the warning information is generated and sent to the management terminal for manual verification and confirmation by the management personnel;

[0018] When the current state is an alarm state, an alarm message is generated and sent to the management terminal and an alarm is sounded to drive away the corresponding personnel.

[0019] In one embodiment, the sending to the administrator terminal via the Internet of Things is implemented as follows:

[0020] Transmit alarm information through the Internet of Things NB-IOT and interact with the management personnel terminal.

[0021] In one embodiment, the water surface life-saving equipment includes a floating carrying unit, a power unit, a control unit and a life buoy. The floating carrying unit floats in the water and carries other parts. A GPS positioning system is installed on it to facilitate the rescue personnel to determine the drowning location; the control unit receives the coordinate information sent by the management terminal, controls the power unit to drive the water surface life-saving equipment to move to the position of the personnel in the reservoir, and the life buoy is fixed to the floating carrying unit by an electromagnetic buckle and connected to the floating carrying unit by a life rope. When the water surface life-saving equipment moves to the set position coordinates, the control unit controls the electromagnetic buckle to cut off power and release the life buoy.

[0022] The second aspect of the present invention further provides a reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm, which is used to implement the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm described in the first aspect, and is characterized by comprising:

[0023] An image capture module is used to continuously capture images of the water area including the edge of the reservoir based on visual sensing technology after a person enters the water monitoring range;

[0024] The data processing training and recognition module is used to clean, annotate and enhance water image data to improve the generalization ability of the model, train the YOLOv5 model, and use the trained model to detect and identify human targets in water images in real time;

[0025] IoT transmission module, used to transmit the judgment results of the YOLOv5 model to the management terminal;

[0026] The alarm module generates alarm information according to the status type and deploys surface life-saving equipment.

[0027] In one embodiment, the image capture module is a panoramic camera device floating on the water surface; the Internet of Things transmission module is an NB-IOT communication module; the training data set used by the data processing training recognition module contains accurately labeled human body images in water areas.

[0028] The second aspect of the present invention also provides a reservoir Internet of Things monitoring and alarm device based on the YOLOv5 algorithm, including:

[0029] Image acquisition equipment, which continuously captures images of the water area including the edge of the reservoir based on visual sensing technology after personnel enter the water monitoring range;

[0030] An electronic device, comprising a memory and a processor, wherein the memory is used to store computer program instructions, and the processor takes the water area image as input and runs the computer program instructions to enable the electronic device to execute the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm described in the first aspect of the present invention;

[0031] The surface lifesaving equipment floats on the water surface and carries the image acquisition equipment, communicates with the electronic equipment or the management personnel terminal, receives the alarm information, and moves to the personnel position coordinates when in the alarm state.

[0032] In one embodiment, the electronic equipment is carried by the water surface lifesaving equipment, and the water surface lifesaving equipment also carries batteries or solar panels to provide power for various power-consuming parts.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. By dividing and defining the monitoring range of the water area, the present invention can not only detect targets more accurately, but also make different alarms according to the conditions in different areas to avoid false detection and missed detection, and also provide a basis for rescue preparation work.

[0035] 2. The present invention realizes reservoir IoT monitoring and alarm based on the YOLOv5 algorithm, which can realize the effective recognition of small targets. Through the recognition of images by the YOLOv5 model rather than the naked eye observation of managers, the occurrence of phenomena such as unauthorized entry into the water can be better detected.

[0036] 3. The present invention can determine the current water state type through images, and issue warnings or even automatically provide rescue according to the current state, thereby improving the safety factor of the reservoir. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 This is a flow chart of the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm provided by the present invention.

[0039] Figure 2 This is a structural diagram of the reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm provided by the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] The purpose of the present invention is to provide a reservoir Internet of Things monitoring and alarm method, system and equipment based on the YOLOv5 algorithm, which can improve the safety factor of the reservoir.

[0042] refer to Figure 1 As shown, the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm of the present invention mainly includes the following steps:

[0043] The water monitoring range is monitored. When it is determined that someone has entered the water monitoring range, the water area images including the edge of the reservoir are continuously captured based on visual sensing technology.

[0044] After resizing and normalizing the obtained water area image, the YOLOv5 algorithm is used to obtain the personnel location information, and the current state type of the water area is determined based on the location information.

[0045] Alarm information is generated according to the state type and sent to the management terminal through the Internet of Things; when it is an alarm state, the surface life-saving equipment is controlled to move to the coordinates of the personnel's position. Thus, the method of the present invention has an automatic life-saving equipment deployment mechanism.

[0046] The present invention defines a water monitoring range and a reservoir edge. For example, the water monitoring range may be a range of 5 meters from the edge of the reservoir surface, and the reservoir edge is a range closer to the edge of the reservoir surface, for example, it may be a range of 2 meters from the edge of the reservoir surface. In the water monitoring range and the reservoir edge, since the personnel have not yet entered the water surface, the detection is for targets on land, and the detection effect is obviously better. With the current detection technology, it is almost impossible to miss or misdetect. Therefore, the method for determining whether a person has entered the water monitoring range may be any current mainstream algorithm. When a person enters the water monitoring range, the algorithm of the present invention is run again to save resources. At this time, the personnel information can be collected through the camera, the personnel can be located using the YOLOv5 algorithm, and the distance between the person and the edge of the reservoir can be analyzed. When the distance is less than 2 meters, it can be determined that someone has entered the edge of the reservoir. At the same time, an infrared sensor can be installed on the surface lifesaving equipment, combined with the detection results of the image sensor, double protection can be performed, and the status of the personnel can be tracked and warned in real time.

[0047] In order to realize the detection of the above-mentioned different ranges, the present invention can also define an electronic area, and the electronic area can be automatically expanded according to the edge of the water surface of the reservoir to obtain a corresponding preset distance.

[0048] The present invention adopts the YOLOv5 algorithm to obtain personnel location information, and the implementation method is as follows:

[0049] The recognition principle of the YOLOv5 algorithm is based on the target detection framework of deep learning. Its core idea is to regard the task of identifying human target detection in water as a regression problem, and directly predict the position and category of human targets in the image through the CNN convolutional neural network.

[0050] After the transmitted image is preprocessed by resizing and normalizing, the preprocessed water area image is first sliced ​​through the fcous layer to reduce the amount of convolution calculation. Then YOLOv5 uses the backbone network architecture to extract the features of the human body in the water area image, including shape, spatial relationship, scale, context and motion action. The obtained features enter the neck module and are fused through the feature pyramid network and the path aggregation network. The neck module can enhance the perception of the size difference caused by the distance difference between the human body and the camera in the water area, and further enhance the feature fusion and positioning capabilities; the fused features enter the Head module for final target prediction; the YOLOv5 model of the present invention adopts an advanced attention mechanism. Specifically, in order to cope with the influence of complex situations such as the sparkling water surface and part of the human body under the water on feature extraction, the original CSP module is replaced with a module containing CBAM (Convolutional Block Attention Module) or ECA (Efficient Channel Attention) in the backbone network architecture. The CBAM module enhances the network's ability to express features by integrating channel attention and spatial attention mechanisms. It first uses global pooling operations and convolutional layers to extract channel and spatial attention weights respectively, and then applies these weights to the input features to strengthen the response to the target features. ECA is an efficient channel attention mechanism that improves the model's sensitivity to different channel features by adaptively adjusting the response of the channel. Through this replacement, the problem of insufficient input feature information and insufficient feature information extraction caused by the small volume of the human body emerging on the water surface can be effectively improved. The Backbone part of YOLOv5 can not only extract richer and more robust features, but also maintain high efficiency, so that the network can capture features more finely. Even when the target part is blocked or only a small part is exposed above the water surface, it can still be effectively identified, improving the efficiency and accuracy of human body recognition in the water, thereby providing strong technical support for the reservoir Internet of Things monitoring and alarm system.

[0051] Furthermore, in order to more accurately extract the characteristics of the human body in the water area, the YOLOv5 algorithm of the present invention simultaneously obtains the relevant position and characteristic information of the human body, and judges the current state type of the water area in combination with the characteristic information. At this time, the YOLOv5 algorithm can be combined with the MediaPipe model, and the MediaPipe model can be used to obtain the key point information of the human body, estimate the posture of the human body, and judge whether drowning occurs. For example, drowning characteristics such as lowering the head, raising the head, opening the mouth, and closing the eyes. Taking the mouth as an example, according to the key point information of the mouth, the length and width of the mouth are calculated, so as to judge whether the mouth is open, the number of times the mouth is opened, etc. If an abnormality is found, the alarm device is triggered in time.

[0052] The states determined by the present invention include normal state, warning state and alarm state. The normal state means that there are no people in the reservoir and at the edge of the reservoir, the warning state means that there are people at the edge of the reservoir but no people in the reservoir, and the alarm state means that there are people in the reservoir, such as swimming or doing other dangerous activities. Furthermore, when the MediaPipe model is combined to determine that there is a drowning phenomenon in the reservoir, the alarm state can also be upgraded to an emergency state.

[0053] Among them, the alarm information is generated according to the status type, and the specific implementation method is as follows:

[0054] When the current status is normal, remain silent and continue monitoring.

[0055] When the current state is the warning state, a warning message is generated and sent to the management terminal for manual verification and confirmation by the management. At this time, the management can control the surface life-saving equipment to move toward the edge of the reservoir where the personnel are located, so as to be ready to start rescue at any time.

[0056] When the current state is an alarm state, an alarm message is generated and sent to the management terminal and an alarm is sounded to drive away the corresponding personnel. At this time, the water surface lifesaving equipment can be controlled by the controller to automatically move to the location of the personnel.

[0057] Furthermore, the method of sending the information to the administrator terminal via the Internet of Things is as follows:

[0058] Transmit alarm information through the Internet of Things NB-IOT and interact with the management personnel terminal.

[0059] The water surface lifesaving equipment of the present invention comprises a floating bearing unit, a power unit, a control unit and a lifebuoy. The control unit has a communication function. The floating bearing unit floats in the water area and carries other parts. A GPS positioning system can be installed on it to facilitate the rescue personnel to determine the drowning position. Furthermore, a grid can be divided on the surface of the reservoir to assist in target positioning, which is of obvious significance to some small and medium-sized reservoirs. In the early warning state or the conventional alarm state, the control unit receives the coordinate information sent by the management terminal, and controls the power unit to drive the water surface lifesaving equipment to move to the edge of the reservoir or the position of the personnel in the reservoir. In an emergency state, the control unit receives the signal triggered by the identification part and the coordinate information automatically sent, and controls the power unit to drive the water surface lifesaving equipment to move to the position of the personnel in the reservoir. The identification part can be a separate processor or integrated in the control unit. The lifebuoy can be fixed to the floating bearing unit through an electromagnetic buckle, and is connected to the floating bearing unit through a life rope. When the water surface lifesaving equipment moves to the set position coordinates, the control unit controls the electromagnetic buckle to cut off power to release the lifebuoy. Through this device, the lifebuoy can be quickly released in an emergency to provide immediate rescue for the person who falls into the water, and the life-saving equipment can be retrieved after the alarm state is lifted.

[0060] In order to execute the above-mentioned reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm to achieve the corresponding functions and technical effects, a reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm is provided below.

[0061] refer to Figure 2 As shown, the system mainly includes:

[0062] The image capture module is used to continuously capture images based on visual sensing technology after a person enters the water monitoring range to obtain an image of the water area including the edge of the reservoir. For example, the image capture module can be a panoramic camera device floating on the water surface.

[0063] The data processing training recognition module is used for cleaning, labeling, and enhancing water image data to improve the generalization ability of the model, train the YOLOv5 model, and use the trained model to detect and recognize human targets in water images in real time. For example, the training data set used by the data processing training recognition module contains a large number of accurately labeled water human images, which provide YOLOv5 with rich learning materials, enabling it to effectively recognize human bodies in water areas.

[0064] The Internet of Things transmission module is used to transmit the YOLOv5 model judgment result to the management terminal. For example, the Internet of Things transmission module can be a NB-IOT communication module.

[0065] The alarm module generates alarm information according to the status type and deploys surface life-saving equipment.

[0066] The reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm provided by the present invention is realized based on visual technology, the YOLOv5 algorithm model and the Internet of Things NB-IOT information transmission.

[0067] Among them, the determination of whether a person has entered the monitoring range of the water area can be achieved through the cameras set up at the edge of the reservoir and the corresponding detection model, which is an existing technology.

[0068] The present invention also provides a reservoir Internet of Things monitoring and alarm device based on the YOLOv5 algorithm, which mainly includes:

[0069] Image acquisition equipment continuously captures images of the water area including the edge of the reservoir based on visual sensing technology after personnel enter the water monitoring range.

[0070] The electronic device includes a memory and a processor, wherein the memory is used to store computer program instructions, and the processor takes the water area image as input and runs the computer program instructions to enable the electronic device to execute the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm of the present invention. In practical applications, the electronic device can be a server.

[0071] The water surface rescue equipment floats on the water surface and carries the image acquisition equipment, communicates with the electronic equipment or the management terminal, receives the alarm information, and moves to the personnel position coordinates when in the alarm state. The electronic equipment can be carried by the water surface rescue equipment, and the water surface rescue equipment also carries batteries or solar panels to provide power for various electrical parts.

[0072] The images captured by the camera are transmitted to the local server of the reservoir, and the YOLOv5 model that has completed big data training is used on the local PC to determine the current status. The alarm information is then transmitted to the reservoir and river basin management personnel terminal or uploaded to the cloud server or local server, providing more intelligent and automated monitoring and alarm functions.

[0073] The equipment of the present invention integrates advanced monitoring and rescue functions, wherein the surface lifesaving equipment has a stable floating structure that can float on the water surface and adapt to the changing water environment, is equipped with a panoramic camera device that can capture all-round water images, and a NB-IOT communication module for cloud platform data transmission. At the same time, it can also be equipped with an efficient power management system, which provides long-lasting power for the monitoring equipment through batteries or solar panels. In addition, the core of the equipment is an electronic device including a memory and a processor, in which advanced computer program algorithm models, especially the YOLOv5 algorithm, are stored, and the processor runs these programs according to the images transmitted by the underwater camera to perform real-time monitoring, state judgment, alarm generation and emergency response functions. The equipment has an automatic lifesaving equipment deployment mechanism, which can quickly release the lifebuoy in an emergency to provide immediate rescue for the drowning person. The design of the entire equipment takes into account the structural stability and maintenance convenience, ensuring long-term stable operation and efficient maintenance.

[0074] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be a personal computer, a notebook computer, a mobile phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device or a combination of the above devices.

[0075] In summary, the system and device of the present invention integrate the innovative technologies of panoramic cameras and floating platforms. By deploying panoramic cameras on floating platforms, all-round visual monitoring of the reservoir surface is achieved, and images of the water area are continuously captured. The system uses the YOLOv5 algorithm to perform in-depth analysis on the collected images. CBAM improves the expression ability of features by combining channel attention and spatial attention. ECA, as a lightweight channel attention mechanism, helps to improve the responsiveness of the network to different channel features. This improvement helps to improve the detection performance of the model for small targets in complex water surface scenes, especially when the target is partially exposed to the water surface or occupies a small proportion in the image due to the long distance. Thereby accurately extracting the position and feature information of the human body and other targets, intelligently judging and classifying the water state into normal, warning or alarm state. When an emergency situation is detected, such as drowning or ship collision, the system will not only generate and send an alarm message to the management personnel, but also automatically start a series of intervention measures, including automatic deployment of life-saving equipment. The system triggers the release of automatic lifebuoys, quickly reaches the designated location, and provides immediate rescue support for the drowning person. In addition, the system can also notify managers in real time through NB-IOT Internet of Things technology, helping them to better understand the on-site situation and formulate rescue plans. The system uses big data analysis model training to simulate real scenes in the water, significantly reducing the number of people drowning in the water due to lack of personnel or insufficient manual management, while improving the monitoring capabilities of special actions such as illegal fishing and waste dumping, comprehensively improving the safety monitoring level of reservoirs, and ensuring the safety and rational use of water resources.

[0076] It should also be noted that the terms "include," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, i.e., a process, method, article, or apparatus comprising a list that includes not only those elements but also those not expressly specified or inherent to the process, method, article, or apparatus. Without further limitation, by stating an element "comprising the elements of ..." does not exclude the inclusion of other identical elements in a process, method, article, or apparatus that are not expressly specified or inherent to the process, method, article, or apparatus.

[0077] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only independent technical solutions. The description of this specification is only for the sake of clarity. Those skilled in the art should treat the specification as a whole. The technical solutions in each implementation method can also be appropriately combined to form other understandable implementation methods.

Claims

1. A reservoir IoT monitoring and alarm method based on YOLOv5 algorithm. It is characterized in that The steps include: When it is determined that someone has entered the water monitoring range, the visual sensing technology is used to continuously capture images of the water area including the edge of the reservoir; After resizing and normalizing the water area image, the YOLOv5 algorithm is used to obtain the personnel location information, and the current state type of the water area is determined according to the location information, including normal state, warning state and alarm state; the normal state means that there are no personnel in the reservoir and at the edge of the reservoir, the warning state means that there are personnel at the edge of the reservoir but no personnel in the reservoir, and the alarm state means that there are personnel in the reservoir; Generate alarm information based on the status type and send it to the management terminal through the Internet of Things; Among them, when it is in the alarm state, the surface lifesaving equipment is controlled to move to the personnel position coordinates.

2. According to the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to claim 1, It is characterized in that The water area monitoring range is 5 meters away from the edge of the reservoir surface, and the reservoir edge is 2 meters away from the edge of the reservoir surface. Personnel information is collected through cameras, and the YOLOv5 algorithm is used to locate personnel and analyze their distance from the edge of the reservoir. When the distance is less than 2 meters, it can be determined that someone has entered the edge of the reservoir. At the same time, infrared sensors are installed on the water surface life-saving equipment, and combined with the detection results of image sensors, the status of personnel can be tracked and warned in real time.

3. According to the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to claim 1, It is characterized in that The YOLOv5 algorithm is used to obtain personnel location information, and the implementation method is as follows: First, the preprocessed water image is sliced ​​through the fcous layer to reduce the amount of convolution calculation. Then YOLOv5 uses the backbone network architecture to extract the features of the human body in the water image, including shape, spatial relationship, scale, context and motion. The obtained features enter the neck module and are fused through the feature pyramid network and path aggregation network; the fused features enter the Head module for final target prediction; in the backbone network architecture, the CSP module is replaced by a module containing CBAM or ECA.

4. According to the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to claim 3, It is characterized in that The YOLOv5 algorithm is combined with the MediaPipe model, and the MediaPipe model is used to obtain the key point information of the human body, estimate the posture of the human body, and determine whether drowning occurs.

5. According to the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to claim 1, It is characterized in that The method for generating alarm information according to the status type is as follows: When the current status is normal, remain silent and continue monitoring; When the current status is the warning status, the warning information is generated and sent to the management terminal for manual verification and confirmation by the management personnel; When the current state is an alarm state, an alarm message is generated and sent to the management terminal and an alarm is sounded to drive away the corresponding personnel.

6. According to the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to claim 1, It is characterized in that The water surface life-saving equipment includes a floating carrying unit, a power unit, a control unit and a life buoy. The floating carrying unit floats in the water and carries other parts, and a GPS positioning system is installed on it; the control unit receives the coordinate information sent by the management personnel terminal, controls the power unit to drive the water surface life-saving equipment to move to the position of the personnel in the reservoir, and the life buoy is fixed to the floating carrying unit by an electromagnetic buckle and connected to the floating carrying unit by a life rope. When the water surface life-saving equipment moves to the set position coordinates, the control unit controls the electromagnetic buckle to cut off power and release the life buoy.

7. A reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm, used to implement the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to any one of claims 1 to 6, It is characterized in that include: An image capture module is used to continuously capture images of the water area including the edge of the reservoir based on visual sensing technology after a person enters the water monitoring range; The data processing training and recognition module is used to clean, annotate and enhance water image data to improve the generalization ability of the model, train the YOLOv5 model, and use the trained model to detect and identify human targets in water images in real time; IoT transmission module, used to transmit the judgment results of the YOLOv5 model to the management terminal; The alarm module generates alarm information according to the status type and deploys surface life-saving equipment.

8. According to the reservoir Internet of Things monitoring and alarm system based on the YOLOv5 algorithm according to claim 7, It is characterized in that The image capture module is a panoramic camera device floating on the water surface; the Internet of Things transmission module is a NB-IOT communication module; the training data set used by the data processing training and recognition module contains accurately labeled human body images in water areas.

9. A reservoir IoT monitoring and alarm device based on YOLOv5 algorithm. It is characterized in that include: Image acquisition equipment, which continuously captures images of the water area including the edge of the reservoir based on visual sensing technology after personnel enter the water monitoring range; An electronic device, comprising a memory and a processor, wherein the memory is used to store computer program instructions, and the processor takes the water area image as input and runs the computer program instructions to enable the electronic device to perform the reservoir Internet of Things monitoring and alarm method based on the YOLOv5 algorithm according to any one of claims 1 to 6; The surface lifesaving equipment floats on the water surface and carries the image acquisition equipment, communicates with the electronic equipment or the management personnel terminal, receives the alarm information, and moves to the personnel position coordinates when in the alarm state.

10. The reservoir IoT monitoring and alarm device based on the YOLOv5 algorithm according to claim 9, It is characterized in that The electronic equipment is carried by the water surface lifesaving equipment, and the water surface lifesaving equipment also carries batteries or solar panels to provide power for various power-consuming parts.

Images