Remote management method and system for intelligent video analysis algorithm rule
By collaboratively managing intelligent video analytics algorithm rules at the edge computing end and the remote server end, the problems of high maintenance workload and high false alarm rate of intelligent video analytics algorithms in the industrial sector are solved, achieving efficient rule maintenance and accurate early warning results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DONGXIN LANGYUN (BEIJING) TECH DEV CO LTD
- Filing Date
- 2026-02-10
- Publication Date
- 2026-05-05
AI Technical Summary
In the industrial sector, intelligent video analytics algorithms require significant maintenance when setting and canceling rules, and have a high false alarm rate, resulting in insufficient applicability of algorithm rules and inaccurate prediction of abnormal events.
Deploy algorithm rules at the edge computing end and manage them through a remote server. Combine the initial identification at the edge computing end with the secondary identification by the AI agent at the remote server to optimize the algorithm rules, reduce the false alarm rate and improve accuracy.
It enables efficient maintenance of algorithm rules and accurate early warning, reduces false alarm rate, and improves the applicability of video surveillance and user trust.
Smart Images
Figure CN121985148A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of video surveillance technology, and more specifically to a remote management method and system for intelligent video analysis algorithm rules. Background Technology
[0002] Currently, video surveillance systems are widely used across various industries. In the industrial sector, many companies have installed and deployed a large number of high-definition cameras. Due to the limited number of security and protection personnel, the massive amounts of data present a significant challenge to video surveillance operations. Faced with real-time video data often measured in terabytes, manual monitoring is not only difficult to detect abnormal videos in a timely manner, but also increasingly challenging to find anomalies in stored historical video data. Intelligent video analysis algorithms can effectively extract abnormal frames from real-time or historical video data according to pre-trained rules. Therefore, they can replace manual monitoring of video data, detect abnormal data, and provide alarms, greatly improving the efficiency of video surveillance data analysis.
[0003] However, in practical applications, there are two problems. First, setting and canceling intelligent video algorithm rules must be done manually, resulting in a large workload for maintaining algorithm rules, and the consistency between the re-set algorithm rules and those before cancellation is poor. Second, as the scene changes, the intelligent video analysis algorithm will generate a high false alarm rate. The alarms generated by a large number of false alarm data will create pressure on alarm cancellation and reduce users' trust in the algorithm alarms.
[0004] Therefore, improving the applicability of intelligent video algorithms in industrial applications and the accuracy of abnormal event warnings is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0005] In view of the above problems, the present invention proposes a remote management method and system for intelligent video analysis algorithm rules, so as to overcome the above problems or at least partially solve the above problems.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] In a first aspect, the present invention provides a remote management method for intelligent video analysis algorithm rules, comprising the following steps: At the edge computing end, a set of algorithm rules for analyzing surveillance videos are deployed for different control areas; The edge computing terminal submits the heartbeat signals and algorithm rule settings of each controlled area to the remote server; The remote server sends corresponding update commands to the edge computing terminal based on the relevant data received from different control areas; The edge computing terminal executes the corresponding operation based on the received remote instructions and sends the execution result back to the remote server. At the edge computing end, algorithms are used to initially identify whether there are any anomalies in the surveillance video, output alarm images, and send them to the remote server. Deploy an AI agent on a remote server to perform secondary recognition on alarm images and generate suggestions on whether to retain or discard them.
[0008] Furthermore, the edge computing interface is divided into multiple control zones, each corresponding to a set of algorithm rules, and each algorithm rule has a unique ID. For each control zone, there are four buttons: Set Rule, Pause Rule, Resume Rule, and Delete Rule. The Set Rule button is used to configure the corresponding algorithm rules for the control zone; the Pause Rule button is used to temporarily disable all configured algorithm rules for the currently selected control zone without deleting the data; the Resume Rule button is used to enable historically saved configurations; and the Delete Rule button is used to delete the configured corresponding algorithm rules.
[0009] Furthermore, the heartbeat signal submitted by the edge computing terminal to the remote server is used to indicate whether the monitoring equipment in each control area is online; the algorithm rule setting status includes whether algorithm rules have been set in each control area, all algorithm rule IDs, currently used algorithm rules, and suspended algorithm rules; the remote server displays the online status of the monitoring equipment in each control area, the list of algorithm rules, and the running status in the management interface.
[0010] Furthermore, the remote server sends a rule pause command or rule resume command to the edge computing terminal based on the algorithm rule ID; When the edge computing terminal receives a rule pause command, it temporarily stops the corresponding algorithm rule from performing video analysis on the corresponding controlled area; when the edge computing terminal receives a rule resume command, it re-activates the previously stored algorithm rule to perform video analysis on the corresponding controlled area.
[0011] Furthermore, when the remote server sends an update command to the edge computing terminal, it simultaneously sends a checksum authentication information. If the edge computing terminal receives an update command that fails the checksum authentication, it discards the current update command; otherwise, it executes the current update command.
[0012] Furthermore, operation logs of algorithm rules are recorded at the edge computing end. The operation logs record the timestamps of each algorithm rule setting, pausing, and resuming operation, whether the operation source is local or remote, the instruction verification bit authentication information digest, and the operation result status.
[0013] Furthermore, after receiving a suggestion to retain the current alarm image, the edge computing terminal determines it as a valid warning; when it receives a suggestion to discard the current alarm image, it determines that there is a false alarm and filters out the current alarm image.
[0014] Furthermore, alarm images that are recommended to be retained and alarm images that are recommended to be discarded are continuously stored in the alarm data pool and tagged; the algorithm rules are continuously trained and optimized based on the alarm images in the alarm data pool.
[0015] Secondly, the present invention provides a remote management system for intelligent video analysis algorithm rules, which adopts the method described above. The system includes an edge computing terminal and a remote server terminal. An intelligent video analytics module is deployed on the edge computing terminal, which includes a rule execution unit, a reporting unit, and a first alarm unit; a monitoring unit and a second alarm unit are deployed on the remote server terminal. The rule enforcement unit deploys a set of algorithm rules for analyzing surveillance videos for different control areas; The reporting unit submits the heartbeat signals and algorithm rule settings of each controlled area to the monitoring unit; Based on the relevant data received from different control areas, the monitoring unit sends corresponding update instructions to the rule execution unit; The rule execution unit performs corresponding operations based on the received remote instructions and feeds back the execution results to the monitoring unit; The first alarm unit uses algorithm rules to initially identify whether there are any abnormalities in the surveillance video, outputs an alarm image, and sends it to the second alarm unit. The second alarm unit uses an AI agent to perform secondary recognition on the alarm images and generates suggestions on whether to retain or discard them.
[0016] Furthermore, the intelligent video analysis module also includes an optimization unit. The optimization unit continuously stores alarm images that are recommended to be retained and alarm images that are recommended to be discarded into the alarm data pool and tags them. Based on the alarm images in the alarm data pool, the algorithm rules are continuously trained and optimized.
[0017] As can be seen from the above technical solution, compared with the prior art, the present invention has the following beneficial effects: 1. This invention configures a set of multiple algorithm rules for video analysis for each controlled area. It can not only be maintained locally, but also managed remotely through a remote server. According to the actual needs of scenario risk management, it can update the algorithm rules that have been initialized, which can effectively improve the maintenance efficiency and quality of algorithm rules.
[0018] 2. This invention first uses algorithmic rules to preliminarily identify abnormal situations in videos and generate early warning data. Then, an AI agent is used to perform secondary identification on the early warning data. Through the collaborative analysis of the video by the two, the accuracy of early warning can be effectively improved and the false alarm rate of algorithmic rules can be reduced. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0020] Figure 1 A flowchart of a remote management method for intelligent video analysis algorithm rules provided in an embodiment of the present invention; Figure 2 This is a flowchart of maintaining algorithm rules provided in an embodiment of the present invention; Figure 3 This is a flowchart of remote monitoring of monitoring equipment provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the collaborative early warning process of algorithm rules and AI agents provided in this embodiment of the invention. Detailed Implementation
[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0022] like Figure 1 As shown in the figure, this invention discloses a remote management method for intelligent video analysis algorithm rules, including the following steps: At the edge computing end, a set of algorithm rules for analyzing surveillance videos are deployed for different control areas; The edge computing terminal submits the heartbeat signals and algorithm rule settings of each controlled area to the remote server; The remote server sends corresponding update commands to the edge computing terminal based on the relevant data received from different control areas; The edge computing terminal executes the corresponding operation based on the received remote instructions and sends the execution result back to the remote server. At the edge computing end, algorithms are used to initially identify whether there are any anomalies in the surveillance video, output alarm images, and send them to the remote server. Deploy an AI agent on a remote server to perform secondary recognition on alarm images and generate suggestions on whether to retain or discard the alarm images. The secondary recognition result serves as the final alarm result.
[0023] In one embodiment, the edge computing interface is divided into multiple control zones, each corresponding to a set of algorithm rules, and each algorithm rule has a unique ID. Each control zone includes four buttons: Set Rule, Pause Rule, Resume Rule, and Delete Rule. The Set Rule button configures the corresponding algorithm rules for the control zone; the Pause Rule button temporarily disables all configured algorithm rules for the currently selected control zone without deleting the data; the Resume Rule button re-enables previously saved configurations; and the Delete Rule button deletes the configured algorithm rules. All operations require secondary confirmation to prevent accidental activation.
[0024] like Figure 3 As shown, the edge computing terminal periodically submits heartbeat signals and algorithm rule settings to the remote server via an HTTP interface. The remote server parses the received data. The heartbeat signal indicates whether the monitoring devices in each controlled area are online. The algorithm rule settings include whether algorithm rules have been set in each controlled area, all algorithm rule IDs, currently used algorithm rules, and paused algorithm rules. The remote server displays the online status, algorithm rule list, and running status of the monitoring devices in each controlled area in the management interface. When the remote server does not receive a heartbeat signal within a timeout period, it marks the monitoring device as offline and triggers an alarm. When a heartbeat signal is received, the last contact time of the monitoring device is updated, the monitoring device is marked as online, the submitted algorithm rule setting data is parsed, and the online status, algorithm rule list, and running status of the monitoring devices corresponding to that controlled area are displayed in the management interface.
[0025] Specifically, such as Figure 2 As shown, the process for performing local and remote operations on algorithm rules is as follows: When performing local operations on algorithm rules, clicking the corresponding button on the edge computing terminal interface will bring up a secondary confirmation dialog box for the user to confirm before executing the local control program. When performing remote operations on algorithm rules, the remote server sends a rule pause command or rule resume command to the edge computing terminal via an HTTP interface based on the algorithm rule ID. Both commands sent by the remote server to the edge computing terminal include verification information to ensure security. Upon receiving the remote command, the edge computing terminal executes a verification process. If the edge computing terminal receives an update command that fails verification, it discards the current update command; otherwise, it executes the current update command and runs the local control program.
[0026] When the edge computing terminal receives a rule pause command, it temporarily stops the corresponding algorithm rule from performing video analysis on the corresponding controlled area; when the edge computing terminal receives a rule resume command, it re-activates the previously stored algorithm rule to perform video analysis on the corresponding controlled area.
[0027] In addition, operation logs of algorithm rules are recorded at the edge computing end. The operation logs record the timestamps of each algorithm rule setting, pausing, and resuming operation, the operation source (local or remote), the instruction check bit authentication information digest, and the operation result status. The logs are stored in the local EPROM or hard disk, and can retain historical data of no less than 30 days. They can also be compared and verified with relevant rule data in the remote server.
[0028] After the edge computing terminal updates the algorithm rules and execution status, it sends the execution results back to the remote server via an HTTP interface, enabling the remote client to synchronously update the algorithm rule status of the corresponding control area.
[0029] Finally, the execution results are sent back to the server via an HTTP interface to update the algorithm rule status.
[0030] In one embodiment, after receiving a suggestion to retain the current alarm image, the edge computing terminal determines it as a valid warning and reports it to the business platform, outputting alarm information. Upon receiving a suggestion to discard the current alarm image, it determines there is a false alarm and filters out the current alarm image. The alarm images recommended for retention and those recommended for discard are continuously stored in an alarm data pool and tagged. The algorithm rules are continuously trained and optimized based on the alarm images in the alarm data pool.
[0031] Specifically, such as Figure 4 As shown, a multimodal AI large model, including the DeepSeek R1 7B and Qwen3 8B AI large models, was deployed privately on a remote server. An intelligent agent was developed, and the process of "receiving images - comparing images - deleting images - submitting filtering opinions" was defined. Relying on the image comparison capability of the large model, the alarm data generated by the intelligent video analysis algorithm unit was filtered, and the filtered alarm images were uniformly saved to the feedback dataset for continuous training and optimization of the algorithm rules.
[0032] The algorithm rules on the edge computing end are deep learning models with video data recognition capabilities formed through deep learning and training. Based on a large amount of historical correct data and false alarm data, it not only continuously improves the ability of the intelligent agent to retain or discard algorithm warning suggestions in terms of accuracy, but also continuously stores the alarm images that are suggested to be retained and those that are suggested to be discarded into the alarm data pool and labels them. It establishes and continuously supplements personalized knowledge graphs for each algorithm rule, realizes continuous training and optimization of algorithm rules, improves the adaptation accuracy of algorithm rules to scenarios, reduces the false alarm rate from the root, and further reduces the algorithm false alarm rate.
[0033] In one embodiment, the present invention also provides a remote management system for intelligent video analysis algorithm rules. Using the above method, the system includes an edge computing terminal and a remote server. An intelligent video analytics module is deployed on the edge computing terminal, which includes a rule execution unit, a reporting unit, and a first alarm unit; a monitoring unit and a second alarm unit are deployed on the remote server terminal. The rule enforcement unit deploys a set of algorithm rules for analyzing surveillance videos for different control areas; The reporting unit submits the heartbeat signals and algorithm rule settings of each controlled area to the monitoring unit; Based on the relevant data received from different control areas, the monitoring unit sends corresponding update instructions to the rule execution unit; The rule execution unit performs corresponding operations based on the received remote instructions and feeds back the execution results to the monitoring unit; The first alarm unit uses algorithm rules to initially identify whether there are any abnormalities in the surveillance video, outputs an alarm image, and sends it to the second alarm unit. The second alarm unit uses an AI agent to perform secondary recognition on the alarm image and generates suggestions on whether to retain or discard the alarm image. The secondary recognition result serves as the final alarm result, and the edge computing terminal issues an alarm based on the secondary recognition result.
[0034] The system of this invention consists of one or more servers and multiple edge computing terminals, and can run on local area networks and wide area networks.
[0035] Even more advantageously, the remote server is also equipped with an optimization unit. This unit continuously stores alarm images that are recommended to be retained and those that are recommended to be discarded into the alarm data pool and tags them. Based on the alarm images in the alarm data pool, the algorithm rules are continuously trained and optimized.
[0036] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0037] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A remote management method for intelligent video analysis algorithm rules, characterized in that, Includes the following steps: At the edge computing end, a set of algorithm rules for analyzing surveillance videos are deployed for different control areas; The edge computing terminal submits the heartbeat signals and algorithm rule settings of each controlled area to the remote server; The remote server sends corresponding update commands to the edge computing terminal based on the relevant data received from different control areas; The edge computing terminal executes the corresponding operation based on the received remote instructions and sends the execution result back to the remote server. At the edge computing end, algorithms are used to initially identify whether there are any anomalies in the surveillance video, output alarm images, and send them to the remote server. Deploy an AI agent on a remote server to perform secondary recognition on alarm images and generate suggestions on whether to retain or discard them.
2. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, The edge computing interface is divided into multiple control zones, each corresponding to a set of algorithm rules, and each algorithm rule has a unique ID. For each control zone, there are four buttons: Set Rule, Pause Rule, Resume Rule, and Delete Rule. The Set Rule button is used to configure the corresponding algorithm rules for the control zone; the Pause Rule button is used to temporarily disable all configured algorithm rules for the currently selected control zone without deleting the data; the Resume Rule button is used to enable historically saved configurations; and the Delete Rule button is used to delete the configured algorithm rules.
3. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, The heartbeat signal submitted by the edge computing terminal to the remote server indicates whether the monitoring equipment in each control area is online; the algorithm rule setting status includes whether algorithm rules have been set in each control area, all algorithm rule IDs, currently used algorithm rules, and suspended algorithm rules; the remote server displays the online status of the monitoring equipment in each control area, the list of algorithm rules, and the running status in the management interface.
4. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, The remote server sends a rule pause command or rule resume command to the edge computing terminal based on the algorithm rule ID; When the edge computing terminal receives a rule pause command, it temporarily stops the corresponding algorithm rule from performing video analysis on the corresponding controlled area. When the edge computing terminal receives the rule restoration instruction, it re-activates the previously stored algorithm rules to perform video analysis on the corresponding controlled area.
5. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, When the remote server sends an update command to the edge computing terminal, it simultaneously sends a checksum authentication information. If the edge computing terminal receives an update command that fails the checksum authentication, it discards the current update command; otherwise, it executes the current update command.
6. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, Operation logs of algorithm rules are recorded at the edge computing end. The operation logs record the timestamps of each algorithm rule setting, pausing, and resuming operation, whether the operation source is local or remote, the instruction verification bit authentication information digest, and the operation result status.
7. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, After receiving a suggestion from the AI agent to retain the current alarm image, the edge computing device determines it to be a valid warning; when it receives a suggestion to discard the current alarm image, it determines that there is a false alarm and filters out the current alarm image.
8. The remote management method for intelligent video analysis algorithm rules as described in claim 1, characterized in that, Alarm images recommended for retention and those recommended for rejection are continuously stored in the alarm data pool and tagged; the algorithm rules are continuously trained and optimized based on the alarm images in the alarm data pool.
9. A remote management system for intelligent video analysis algorithm rules, characterized in that, The system, using the method described in any one of claims 1-8, includes an edge computing terminal and a remote server. An intelligent video analytics module is deployed on the edge computing terminal, which includes a rule execution unit, a reporting unit, and a first alarm unit; a monitoring unit and a second alarm unit are deployed on the remote server terminal. The rule enforcement unit deploys a set of algorithm rules for analyzing surveillance videos for different control areas; The reporting unit submits the heartbeat signals and algorithm rule settings of each controlled area to the monitoring unit; Based on the relevant data received from different control areas, the monitoring unit sends corresponding update instructions to the rule execution unit; The rule execution unit performs corresponding operations based on the received remote instructions and feeds back the execution results to the monitoring unit; The first alarm unit uses algorithm rules to initially identify whether there are any abnormalities in the surveillance video, outputs an alarm image, and sends it to the second alarm unit. The second alarm unit uses an AI agent to perform secondary recognition on the alarm images and generates suggestions on whether to retain or discard them.
10. The remote management system for the intelligent video analysis algorithm rules as described in claim 9, characterized in that, The intelligent video analysis module also includes an optimization unit, which continuously stores alarm images that are recommended to be retained and alarm images that are recommended to be discarded into the alarm data pool and adds tags to them. Based on the alarm images in the alarm data pool, the algorithm rules are continuously trained and optimized.