A video monitoring method and system based on Internet of Things

Abstract

The invention relates to the technical field of Internet of Things video monitoring, and discloses a video monitoring method and system based on the Internet of Things, comprising: step one, dividing #imgabs0# target areas, interconnecting monitoring devices of different models and types in all target areas with a monitoring system, and classifying and collecting video monitoring data by the monitoring system; step two, functionally classifying the monitoring devices, and splicing the coverage of all monitoring angles with each other, and analyzing and discovering the monitoring blind areas that cannot be covered; step three, standardizing the video monitoring data, and then substituting it into a deep learning model framework to train and obtain a standard algorithm model; step four, the standard algorithm model extracts features according to the video monitoring data, performs real-time analysis and recognition with high accuracy, and performs storage and backup in chronological order, and the data storage and backup have strong traceability; step five, the monitoring system judges and generates corresponding reports, and controls the functional parameters of the monitoring device accordingly, and has strong applicable expansion capability and fast response speed.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things video monitoring, and in particular to a video monitoring method and system based on the Internet of Things. Background Art

[0002] With the popularization of 5G networks and the development of artificial intelligence technology, the application of IoT technology will be more extensive. IoT technology connects various types of devices and sensors to the Internet to realize real-time data collection, transmission and analysis. All connected devices can communicate with each other through unified conversion standards and can automatically perform tasks, which greatly improves work efficiency. In the IoT ecosystem, data conversion is usually completed by middleware, platforms or services that interact with IoT devices. Data conversion involves converting raw data from one format or structure to another format or structure to make it more suitable for further processing, analysis or storage. In IoT applications, data conversion operations include but are not limited to unit conversion, data formatting, data cleaning, data aggregation, data mapping, data encoding, data compression, feature extraction or standardization operations. Unit conversion converts the data collected by the sensor from the original unit to a unit of measurement that is convenient for analysis and calculation. Data formatting converts data from its original form to a structured format that is convenient for storage and transmission. Data cleaning operations ensure data quality by removing noise data, correcting errors and eliminating redundancy. Data aggregation operations usually combine multiple data points into a summary data point. Data mapping operations map data from one domain to another, mainly applicable to mathematical transformations. Protocol conversion converts data from one communication protocol to another so that it is compatible with different systems or networks. The purpose of data encoding and data compression operations is to transmit or store data safely and efficiently. Feature extraction extracts representative features from raw data, providing data support for subsequent data analysis or machine learning algorithms. Standardized operations are a key step in realizing data processing and analysis in IoT systems. IoT systems are widely used and popularized in all aspects of smart life. In the construction of smart cities, IoT video surveillance uses AI technology to perform image recognition and behavior analysis on video data, which can realize automatic detection and early warning of abnormal events. This intelligent data processing and analysis capability is not available in traditional video surveillance, which greatly improves the application value and practicality of the monitoring system.

[0003] At present, the traditional IoT-based video surveillance system cannot identify key information in the video screen in real time, and requires manual observation and analysis of recorded video data, which seriously affects the efficiency of data analysis and delays the response time. In addition, the video surveillance device continuously records and uploads data streams during use, which not only consumes a large amount of network bandwidth, but also increases the working pressure of storage devices. Faced with large-scale monitoring needs, the system has blind spots in its scope of application and cannot support the expansion of diversified monitoring equipment. It has poor scalability and is prone to form information islands. Subsequent analysis and decision-making deviations are large, resulting in the loss of precious time to respond to emergencies. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a video surveillance method and system based on the Internet of Things, which has the advantages of high accuracy in real-time analysis and identification, strong applicability and expansion capability, fast response speed, strong traceability of data storage and backup, etc., and solves the problems that traditional video surveillance systems based on the Internet of Things cannot identify key information in real time, have high energy consumption for data storage and backup, poor applicability and expansion capability, and slow response speed.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solution: a video monitoring method based on the Internet of Things, comprising the following steps:

[0008] Step 1: divide Z target areas according to the video surveillance range, interconnect all monitoring devices of different models and types in all target areas with the monitoring system based on the video Internet of Things, and the monitoring system based on the video Internet of Things collects video surveillance data according to different data sources;

[0009] Step 2: Classify the monitoring devices of different models and types according to their functions, and combine them to get the coverage of all monitoring angles, and analyze and find the monitoring blind spots that cannot be covered;

[0010] Step 3: Standardize the video surveillance data collected by monitoring devices of different models and types, and then substitute the standardized video surveillance data into the deep learning model framework to train and obtain a standard algorithm model;

[0011] Step 4: The standard algorithm model extracts features from the video surveillance data collected by different types of surveillance devices, analyzes key information in real time, and stores and backs up the data in chronological order;

[0012] Step 5: The monitoring system based on the video Internet of Things generates corresponding management reports or monitoring reports based on key information judgments, and controls the functional parameters of monitoring devices of different models and types in all areas accordingly.

[0013] A video monitoring system based on the Internet of Things, including an Internet of Things monitoring module, a video analysis module and a monitoring management module;

[0014] The IoT monitoring module includes a regional data unit, a functional data unit and a video data unit. The regional data unit collects a regional data set through a network connection to a database. The regional data set includes the target area and the distribution data of all monitoring devices in the target area. The functional data unit collects a functional data set through a network connection to a monitoring device. The functional data set includes the viewing angle range data, picture function data, voice function data and target tracking function data of the monitoring device. The video data unit collects a video data set through a network connection to a monitoring device. The video data set includes the video picture collected in real time by the monitoring device and the collection time point. The IoT monitoring module transmits the regional data set, the functional data set and the video data set to the video analysis module through the network.

[0015] The video analysis module includes a coverage analysis unit, a deep learning unit and a real-time analysis unit. The coverage analysis unit calculates and generates a coverage ratio Fgbl according to the regional data set and the functional data set, and obtains the total coverage area FM of all monitoring angles in the target area and the blind area MM not covered by the monitoring. The deep learning unit performs standardization processing according to the video data set, removes the picture noise and redundant data in the video data set, and corrects the erroneous data. The deep learning unit is provided with a Transformer deep learning model framework, and substitutes the standardized video data set into the Transformer deep learning model framework, trains to obtain a standard algorithm model BZS, and applies it to the analysis and processing of the video data of all monitoring devices in the target area. The real-time analysis unit substitutes the video data set into the standard algorithm model BZS, analyzes and extracts the key features Gjtz, and transmits them to the database through the network for storage and backup. The video analysis module transmits the coverage ratio Fgbl, the standard algorithm model BZS and the key features Gjtz to the monitoring management module through the network;

[0016] The monitoring management module compares the coverage ratio Fgbl with the coverage threshold, determines the coverage degree of the monitoring device in the target area, and generates a corresponding management report to prompt the management personnel to perform maintenance management. The monitoring management module is set with a monitoring standard. The monitoring management module determines whether the key feature Gjtz has abnormal features according to the monitoring standard through the standard algorithm model BZS, and generates a corresponding monitoring report to control and adjust the functional parameters of the monitoring device.

[0017] Preferably, the expression of the regional data set is f represents the distribution data of all monitoring devices in the target area, to The area corresponding to the 1st to the Zth target areas and the distribution data of all monitoring devices in the target areas.

[0018] Preferably, the expression of the functional data set is {G1, G2, G3, ..., G n}, G1 to G n The subscripts 1 to n correspond to the viewing range data, picture function data, voice function data and target tracking function data of each monitoring device in all target areas in sequence, and the subscripts 1 to n indicate that the data set consists of n function data.

[0019] Preferably, the expression of the video data set is t represents the time point of real-time data collection by the monitoring device, to The digital numbers in the functional data set and the digital numbers in the video data set correspond to each other, and the G1 in the functional data set and the G2 in the video data set correspond to each other. All correspond to the same monitoring device.

[0020] Preferably, the calculation process of the coverage ratio Fgbl is as follows:

[0021] According to the regional data set, the area of ​​the i-th target area and the distribution data of all monitoring devices in the target area are counted, denoted as P i ,in, P i ={PM, PF}, PM represents the area of ​​the i-th target area, PF represents the number and location of all monitoring devices in the i-th target area;

[0022] FM=PF*G n

[0023] MM=PM-FM

[0024]

[0025] In the formula, Fgbl represents the coverage ratio, PF*G n The distribution number and distribution position of all monitoring devices in the i-th target area are combined with the viewing angle range data of the monitoring devices in the functional data set to obtain the total coverage area FM of all monitoring viewing angles in the target area. P-FM represents the target area minus the total coverage area of ​​all monitoring viewing angles to obtain the blind area area MM not covered by the monitoring. The percentage of the blind area not covered by monitoring to the area of ​​the i-th target area is the coverage ratio.

[0026] Preferably, the training process of the standard algorithm model BZS is as follows:

[0027] According to the video data set, the regular video images of each monitoring device in all areas at the same time point are extracted and recorded as CS, where Then, a Gaussian filter is used to remove the picture noise, redundant data and correct the erroneous data in the regular video picture CS of each monitoring device in all areas at the same time point, which is recorded as BCS, where BCS = {BCS1, BCS2, BCS3, ..., BCS n}, BCS1 to BCS n Corresponding to the normal video images of each monitoring device in all areas at the same time point after the standardized processing;

[0028]

[0029] In the formula, BZS represents the standard algorithm model, ∑ Transformer (BCS1+BCS2+BCS3+...+BCS n ) means that the normal video images of each monitoring device in all areas at the same time point that have been standardized are substituted into the Transformer deep learning model framework to train a standard algorithm model.

[0030] Preferably, the key feature Gjtz calculation formula is as follows:

[0031]

[0032] In the formula, Gjtz represents the key feature, It means that the standard algorithm model is used to compare the differences between the video images collected by the monitoring device in real time in the video data set and the regular video images of each monitoring device in all areas at the same time point that have been standardized, and the video images with differences in the video data set are extracted, which are the key features.

[0033] Preferably, the coverage threshold is set to 0-10%. If the coverage ratio Fgbl is higher than the coverage threshold, the monitoring management module determines that the coverage of the monitoring device in the target area is low, and generates a blind spot management report to prompt the management personnel to perform maintenance management.

[0034] Preferably, the monitoring standards include abnormal behavior standards for people and abnormal target standards. If the standard algorithm model BZS determines that there are abnormal features in the key feature Gjtz according to the monitoring standards, an abnormal target report is generated, and the monitoring devices in the area are controlled to turn on the target tracking function, voice function and picture function to adjust the viewing angle range.

[0035] Compared with the prior art, the present invention provides a video monitoring method and system based on the Internet of Things, which has the following beneficial effects:

[0036] 1. The present invention collects regional data sets, functional data sets and video data sets respectively through the Internet of Things monitoring module. The video analysis module is provided with a coverage analysis unit, a deep learning unit and a real-time analysis unit. The coverage analysis unit calculates and generates a coverage ratio Fgbl according to the regional data set and the functional data set, splices the coverage range of all monitoring angles, analyzes and finds monitoring blind spots that cannot be covered, effectively improves the monitoring coverage rate, and enhances the safety protection level of the target area. The deep learning unit standardizes the video data set, and then substitutes it into the deep learning model framework to train and obtain a standard algorithm model BZS, thereby ensuring the generalization ability of the model to process the video data of the monitoring device in real time. The real-time analysis unit extracts features through the standard algorithm model, analyzes and extracts key features Gjtz, analyzes key information in real time, and has high accuracy in real-time analysis and recognition.

[0037] 2. The present invention transmits the key feature Gjtz to the database through a real-time analysis unit, and stores and backs up the data in chronological order to avoid missing key information, thereby ensuring the effectiveness of the decision-making of the monitoring system. The data storage and backup have strong traceability, and the subsequent data retrieval efficiency is higher. It can quickly locate key data and improve the emergency response capability of the overall system.

[0038] 3. The present invention compares the coverage ratio Fgbl with the coverage threshold through the monitoring management module to determine the coverage degree of the monitoring device in the target area. The coverage threshold is set to 0-10%. If the coverage ratio Fgbl is higher than the coverage threshold, the monitoring management module determines that the coverage degree of the monitoring device in the target area is low, and generates a blind spot management report to prompt the management personnel to perform maintenance management. The monitoring management module is set with monitoring standards, and the monitoring standards include abnormal behavior standards and abnormal target standards. The monitoring management module determines whether the key feature Gjtz has abnormal features through the standard algorithm model BZS according to the monitoring standards. If the standard algorithm model BZS determines that the key feature Gjtz has abnormal features according to the monitoring standards, it generates an abnormal target report, and controls the monitoring device in the area to start the target tracking function, the voice function and the picture function to adjust the viewing angle range. It has strong expansion capability and fast response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a step diagram of the method of the present invention;

[0040] Figure 2 It is a flow chart of the system of the present invention. DETAILED DESCRIPTION

[0041] 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.

[0042] See also Figure 1 , a video monitoring method based on the Internet of Things, comprising the following steps:

[0043] Step 1: Divide the target area into Z areas according to the video surveillance range, and interconnect the monitoring devices of different models and types in all target areas with the monitoring system based on the video Internet of Things, which solves the information island problem caused by the differences in device models and data types between traditional monitoring devices. The monitoring system based on the video Internet of Things collects video surveillance data according to different data sources;

[0044] Step 2: Classify the monitoring devices of different models and types according to their functions, and splice them together to get the coverage of all monitoring angles. Analyze and find the monitoring blind spots that cannot be covered, effectively improve the monitoring coverage rate, and enhance the security protection level of the target area;

[0045] Step 3: Standardize the video surveillance data collected by different types of surveillance devices, extract the regular video images of each surveillance device in all areas at the same time point, and then substitute the standardized video surveillance data into the deep learning model framework to train and obtain a standard algorithm model. The obtained standard algorithm model has better generalization ability, which is convenient for quickly judging abnormal data in the subsequent video data analysis process;

[0046] Step 4: The standard algorithm model extracts features from the video surveillance data collected by monitoring devices of different models and types, analyzes key information in real time, and has high accuracy in real-time analysis and recognition. It also stores and backs up data in chronological order. The data storage and backup have strong traceability, and subsequent data retrieval is more efficient. It can quickly locate key data, which improves the emergency response capability of the overall system.

[0047] Step 5. The monitoring system based on the video Internet of Things generates corresponding management reports or monitoring reports based on key information judgments, and controls the functional parameters of monitoring devices of different models and types in all areas accordingly. It has strong expansion capabilities and fast response speed. When faced with anomalies in the video screen, it can promptly detect abnormal behaviors and issue reports to ensure that the target is not lost, greatly improving the effectiveness of security monitoring.

[0048] See also Figure 2,The above method is implemented based on a video surveillance system based on the Internet of Things, which includes an IoT monitoring module, a video analysis module and a monitoring management module;

[0049] The IoT monitoring module includes a regional data unit, a functional data unit, and a video data unit. The regional data unit collects regional data sets through a network connection database. The regional data sets include the target area and the distribution data of all monitoring devices in the target area. The expression of the regional data set is: f represents the distribution data of all monitoring devices in the target area, including the number and location of the monitoring devices. to The area areas corresponding to the 1st to the Zth target areas and the distribution data of all monitoring devices in the target areas are sequentially obtained, so as to facilitate the subsequent rapid determination of the coverage of the monitoring devices in the target areas.

[0050] The function data unit collects the function data set through the network connection monitoring device. The function data set includes the viewing range data, picture function data, voice function data and target tracking function data of the monitoring device. The expression of the function data set is {G1, G2, G3, ..., G n}, G1 to G n The subscripts 1 to n correspond to the viewing range data, picture function data, voice function data and target tracking function data of each monitoring device in all target areas in turn. The data set consists of n functional data. The functional data of all different types of monitoring devices in the area can be clearly controlled, which is conducive to optimizing resource allocation and enhancing the overall system troubleshooting capabilities.

[0051] The video data unit collects video data sets through a network connection monitoring device. The video data sets include the video images collected by the monitoring device in real time and the collection time points. The expression of the video data set is: t represents the time point of real-time data collection by the monitoring device, to The video images and collection time points collected by each monitoring device in all target areas are sequentially corresponded. The data analysis has strong traceability and improves the emergency response capability. The digital numbers in the functional data set and the digital numbers in the video data set correspond to each other. G1 in the functional data set and G2 in the video data set correspond to each other. All correspond to the same monitoring device.

[0052] The IoT monitoring module transmits the regional data set, the functional data set and the video data set to the video analysis module through the network;

[0053] The video analysis module includes a coverage analysis unit, a deep learning unit, and a real-time analysis unit. The coverage analysis unit calculates and generates the coverage ratio Fgbl based on the regional data set and the functional data set, and obtains the total coverage area FM of all monitoring angles in the target area and the blind area MM not covered by the monitoring. The calculation process is as follows:

[0054] According to the regional data set, the area of ​​the i-th target area and the distribution data of all monitoring devices in the target area are counted, denoted as P i ,in, P i ={PM, PF}, PM represents the area of ​​the i-th target area, PF represents the number and location of all monitoring devices in the i-th target area;

[0055] FM=PF*G n

[0056] MM=PM-FM

[0057]

[0058] In the formula, Fgbl represents the coverage ratio, PF*G n The distribution number and distribution position of all monitoring devices in the i-th target area are combined with the viewing angle range data of the monitoring devices in the functional data set to obtain the total coverage area FM of all monitoring viewing angles in the target area. PM-FM represents the target area minus the total coverage area of ​​all monitoring viewing angles to obtain the blind area area MM not covered by the monitoring. The coverage ratio is the percentage of the blind area not covered by monitoring to the area of ​​the i-th target area. According to the coverage ratio Fgbl, the coverage area and blind area of ​​the current monitoring range can be clearly defined, which is convenient for subsequent optimization of monitoring resource allocation and improvement of the safety protection level of the target area.

[0059] The deep learning unit performs standardization processing on the video data set, removes the picture noise and redundant data in the video data set, and corrects the erroneous data. The deep learning unit is equipped with a Transformer deep learning model framework, and substitutes the standardized video data set into the Transformer deep learning model framework, trains and obtains the standard algorithm model BZS, which is applied to the analysis and processing of the video data of all monitoring devices in the target area. The training process is as follows:

[0060] According to the video data set, the regular video images of each monitoring device in all areas at the same time point are extracted and recorded as CS, where Then, a Gaussian filter is used to remove the picture noise, redundant data and correct the erroneous data in the regular video picture CS of each monitoring device in all areas at the same time point, which is recorded as BCS, where BCS = {BCS1, BCS2, BCS3, ..., BCS n}, BCS1 to BCS n Corresponding to the normal video images of each monitoring device in all areas at the same time point after the standardized processing;

[0061] BZS=∑ Transformer (BCS1+BCS2+BCS3+...+BCS n )

[0062] In the formula, BZS represents the standard algorithm model, ∑ Transformer (BCS1+BCS2+BCS3+...+BCS n ) means that the normal video images of each monitoring device in all areas at the same time point that have been standardized are substituted into the Transformer deep learning model framework, and the standard algorithm model is trained. According to the standard algorithm model BZS, the generalization ability of the model to process the video data of the monitoring device in real time is guaranteed.

[0063] The real-time analysis unit substitutes the video data set into the standard algorithm model BZS, analyzes and extracts the key features Gjtz, and transmits them to the database through the network for storage and backup. The calculation formula is as follows:

[0064]

[0065] In the formula, Gjtz represents the key feature, It means that the standard algorithm model is used to compare the differences between the video images collected by the monitoring device in real time in the video data set and the regular video images of each monitoring device in all areas at the same time point that have been standardized, and the video images with differences in the video data set are extracted, which are the key features. According to the key feature Gjtz, the key information is avoided from being missed, and the decision-making effectiveness of the monitoring system is guaranteed.

[0066] The video analysis module transmits the coverage ratio Fgbl, the standard algorithm model BZS and the key feature Gjtz to the monitoring management module through the network;

[0067] The monitoring management module compares the coverage ratio Fgbl with the coverage threshold to determine the coverage degree of the monitoring device in the target area. The coverage threshold is set to 0-10%. If the coverage ratio Fgbl is higher than the coverage threshold, the monitoring management module determines that the coverage degree of the monitoring device in the target area is low, and generates a blind spot management report to prompt the management personnel to perform maintenance management. The monitoring management module is set with monitoring standards, which include abnormal behavior standards and abnormal target standards. The monitoring management module determines whether the key feature Gjtz has abnormal features through the standard algorithm model BZS according to the monitoring standards. If the standard algorithm model BZS determines that the key feature Gjtz has abnormal features according to the monitoring standards, it generates an abnormal target report and controls the monitoring devices in the area to turn on the target tracking function, voice function and picture function to adjust the viewing angle range.

[0068] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A video surveillance system based on the Internet of Things, characterized in that: Including IoT monitoring module, video analysis module and monitoring management module; The IoT monitoring module includes a regional data unit, a functional data unit and a video data unit. The regional data unit is connected to a database via a network to obtain a regional data set. The regional data set includes The distribution data of the target area, the expression of the regional data set is , to The first to the Distribution data of monitoring devices within a target area; The functional data unit is connected to the monitoring device via a network to obtain a functional data set, wherein the functional data set includes functional data of the monitoring device, and the expression of the functional data set is: , to The first to the Functional data of each monitoring device, including viewing angle range; The video data unit is connected to the monitoring device through a network to obtain a video data set, and the video data set includes video images collected in real time by the monitoring device. The expression of the video data set is: , to The first to the The video images collected by the monitoring device in real time, Indicates the time point at which the monitoring device collects the video image. The digital number in the function data set corresponds to the digital number in the video data set. and video datasets All correspond to the same monitoring device; The IoT monitoring module transmits the regional data set, the functional data set and the video data set to the video analysis module via the network; The video analysis module includes a coverage analysis unit, a deep learning unit and a real-time analysis unit. The coverage analysis unit calculates and generates a coverage ratio based on the regional data set and the functional data set. , get the total coverage area of ​​the monitoring view And the blind area not covered by monitoring , the calculation process is as follows: Based on the regional data set, extract the The distribution data of the target area is recorded as ,in, , , Indicates The area of ​​the target area, Indicates The distribution number of all monitoring devices in the target area; In the formula, Indicates the coverage ratio, Indicates The distribution number of monitoring devices in the target area is combined with the viewing angle range of the monitoring devices in the functional data set to obtain the total coverage area of ​​the monitoring viewing angle. , The target area minus the total coverage area of ​​the monitoring view is the blind area not covered by the monitoring. , Indicates that the blind area not covered by monitoring accounts for The percentage of the area of ​​the target area is the coverage ratio; The deep learning unit performs standardization processing according to the video data set, and the standardization processing includes removing the picture noise and redundant data in the video data set, and correcting the erroneous data. The deep learning unit is provided with a Transformer deep learning model framework, and substitutes the standardized video data set into the Transformer deep learning model framework to train and obtain a standard algorithm model. , the standard algorithm model will be trained The method is applied to the analysis and processing of video data of all monitoring devices in the target area, and the training process is as follows: According to the video data set, the regular video images of each monitoring device at the same time point are extracted and recorded as ,in ; Then use Gaussian filter to remove the regular video image The picture noise and redundant data in, as well as the corrected error data, are recorded as ,in, , to For the 1st to the Video images obtained by standardizing the conventional video images of the monitoring devices at the same time point; Substitute the normal video images of each monitoring device into the Transformer deep learning model framework after standardization, and train the standard algorithm model ; The real-time analysis unit substitutes the video data set into the standard algorithm model , analyze and extract key features , through the standard algorithm model , analyze the difference between the real-time video screen and the video screen after the normal processing of each monitoring device, extract the real-time video screen with differences, and record it as the key feature and transmitted to the database through the network for storage and backup; The video analysis module uses the network to , Standard Algorithm Model and key features Transmit to the monitoring management module; The monitoring management module will cover the proportion Compare the coverage threshold, determine the coverage degree of the monitoring device in the target area, and generate a corresponding management report to prompt the management personnel to perform maintenance management. The monitoring management module sets the monitoring standard, and the monitoring management module determines the key features according to the monitoring standard. Check whether there is any abnormality and generate corresponding monitoring report to control and adjust the functional parameters of the monitoring device.

2. The video surveillance system based on the Internet of Things according to claim 1, characterized in that: The coverage threshold is set to 0-10%. If the coverage is higher than the coverage threshold, the monitoring management module determines that the coverage of the monitoring device in the target area is low, and generates a blind spot management report to prompt the management personnel to perform maintenance management. The monitoring standards include abnormal behavior standards and abnormal target standards. If the key feature If there are abnormal features in the area, an abnormal target report is generated, and the monitoring devices in the area are controlled to turn on the target tracking function and voice function.

3. A video monitoring method based on the Internet of Things, applied to a video monitoring system based on the Internet of Things as described in any one of claims 1-2, characterized in that: The following steps are involved: Step 1: Divide by video surveillance range A target area is created, and all monitoring devices in the target area are interconnected with the monitoring system based on the video Internet of Things. The IoT monitoring module obtains the regional data set, the functional data set, and the video data set accordingly. Step 2: Calculate the total coverage area of ​​the monitoring perspective based on the regional dataset and the functional dataset , Blind area not covered by monitoring and coverage ratio ; Step 3: Extract the regular video images of each monitoring device at the same time point based on the video data set, standardize the video images collected by the monitoring device, and then substitute the standardized video images into the Transformer deep learning model framework to train the standard algorithm model. ; Step 4: Using the standard algorithm model Extract features from real-time video images collected by monitoring devices, analyze key information in real time, and generate corresponding key features , and store and back up in chronological order; Step 5: Set coverage thresholds and monitoring standards, and set the coverage ratio Compare the coverage threshold to determine the coverage degree of the monitoring device in the target area, generate corresponding management reports, prompt the management personnel to perform maintenance management, and then determine the key features according to the monitoring standards. Check whether there is any abnormality, generate corresponding monitoring report, and adjust the functional parameters of the monitoring device.

Images